An Overview of Applications and Insights (2024)

1 Introduction

Within the past decade or so, China has made great achievements in the construction of highwaybridges, with the maximum span breaking through and the number of bridges increasing. By the endof 2023, there were 1,079,300 existing highway bridges in China. At the same time, a large numberof bridge structures have entered the middle and late stages of service one after another, and safetyaccidents are frequent, causing widespread concern in society. Optimizing the operation andmaintenance of existing bridges to maintain and improve their safety performance and servicestatus has become an urgent need for the society.

Highway bridge operation and maintenance enhance structural safety and efficiencythrough technological innovation and management strategies, realizing sensor monitoring³¹, ⁷⁰, dronedetection, algorithmic optimization of disease identification, scientific evaluation methods⁷⁶, ¹³, andmultilevel maintenance decision-making system³⁰, ⁶, which lays the foundation for the intelligentdevelopment of the industry. However, there are many problems, such as poor stability of bridgeperception data and high data processing delay. The technical evaluation dimension is single, andthe model accuracy is low. Operation and maintenance decision-making lack of data support, low levelof intelligence. Bridge operation and maintenance intelligence is still in the primary stage, needto strengthen the state perception, evaluation system, decision support and integration withartificial intelligence technology.

The problems faced by bridge O&M urgently require LLM-based agents technology to solve them.Agents, as human-like intelligent entities, are capable of perception, planning, decision-makingand action autonomously⁹⁶, ²⁶.Although limited to specific tasks such as Go games or book retrieval inthe early days, the Transformer architecture introduced by ⁹⁰ has greatly boosted thedevelopment of Language Models. Large Language Models such as OpenAI’s ChatGPT and Google Bert’shave brought new hope to the study of agents by significantly improving comprehension, summarization,reasoning, and language processing with their excellent generalization capabilities. These LargeLanguage Models are now regarded as the “brain” of agents, and have great potential for applicationin the field of bridge operation and maintenance.

In the context of the current rapid development of the research field of Large Language Models andAgents, the emergence rate of new Large Models and Agents has increased significantly. The academiccommunity has paid great attention to this, and the publication of related research results has showna sharp increase. This shows that AI technology is gradually penetrating and transforming variousindustries. Therefore, it is necessary to explore and introduce the corresponding advanced technologyin the field of bridge operation and maintenance in order to promote the innovative development ofthis field.

At the current research stage, agents are more mature in perception, planning, decision-making, andaction. However, the bridge O&M domain is not sufficiently automated and intelligent in terms ofdisposition sensing, data prediction, performance evaluation, decision constraints, emergencyresponse, and disaster mitigation. Therefore, the new generation bridge O&M management systemshould integrate LLM-based agent technology to equip each bridge with a personal assistant. Atthe same time, the system should consider the complementary and independent relationships ofbridge clusters at the road network level to realize synergy, cooperation and competition amongagents. The purpose of this paper is to discuss how agents can optimize bridge O&M tasks, how tobuild a bridge O&M intelligent system, and to analyze the impact of AI on the future development ofthe bridge O&M field, as well as the opportunities and challenges that may be encountered with thedevelopment of agents.

2 Background

In this section, we provide a detailed overview of the evolution of bridge operation and maintenance and LLM-based agentsresearch. We find that the shortcomings of modern bridge health monitoring techniques match theadvantages of LLM-based agents. By analyzing the development trajectories of both, we propose theuse of LLM-based agents to invigorate the field of bridge operation and maintenance and to promote industrial innovationin the current productivity context.

2.2 Overview of LLM-based Agents Research

In order to realize AI as a substitute for human labor, LLM-based agents must possess two keyconditions: first, they should possess advanced comprehension capabilities covering in-depthunderstanding of written language, speech, images, and video, as well as the ability to accuratelyinterpret human intentions and generate autonomous feedback. Second, LLM-based agents should havethe ability to effectively invoke tools or devices, which involves not only direct human manipulationof tools, but more advanced LLM-based agents should be able to perceive the physical environmentthrough sensors and independently mobilize appropriate tools for real-world self-regulation. Thissection next explores in detail current human exploration and progress in these two areas.

2.2.1 Larger Language Models

Language modeling (LM) is the most important task in the field of Natural Language Processing,which is essentially a synthesis of Natural Language Understanding and Natural LanguageGeneration⁸.By enabling computers to learn a large amount of human language data and applyingspecific methods to predict the conditional probabilities between individual tokens (tokens in thispaper take words in a sentence as an example) in a language, we are able to characterize the logicalrelationships between tokens, thus enabling the language model to understand and generate fluentnatural language⁶².Realizing machines with human-like writing and communication capabilities has beena long-term goal pursued by human beings. In academia, the development of language models is usuallydivided into four main stages⁶⁴:

The first stage is the statistical language modeling (SLM) stage. In this phase, the representationof words relies on solo heat vectors characterized by a sparse set of vectors representing words⁸⁹.Although this approach fails to reflect logical relations in the representation of word vectors, itprovides an intuitive and understandable way of characterization. As for the characterization ofsentences, Markov’s assumption is used to reveal the logical connections between words in a sentencethrough conditional probabilities. Specifically, this model fixes the context length, which is alsocalled n-gram language model, among which the bigram and trigram models are the most common¹¹, ⁴⁵. Thisphase of language modeling is mainly used for information retrieval⁴⁷, ³³, ¹⁰¹. However, the main challengefaced by statistical language models is the “curse of dimensionality”, i.e., as the vocabularyincreases, the parameter space required by the model grows exponentially. To alleviate this problem,techniques such as smoothing strategies are widely used to mitigate the effects of data sparsity¹⁵, ⁷².

The second stage is the Neural Language Model (NLM) stage. Researchers have utilized the word2vecmethod to map words to a low-rank vector space , thus effectively overcoming the sparsity problemin the statistical language model¹⁷, ⁵³, ⁴⁹.This approach not only enables word vectors to characterizelogical relations, but also allows intuitive mathematical connections between vectors. Meanwhile,through neural network techniques, such as multilayer perceptual machines⁴⁹ and recurrent neural networks⁵⁸, ⁷¹, ²⁷,researchers are able to learn from a large amount of natural language data toobtain vector representations of words. This distributed word representation approach hasdemonstrated excellent performance in several downstream tasks of natural language processing (NLP),such as machine translation, marking a major breakthrough in the field of NLP in terms ofrepresentation learning and having a far-reaching impact on subsequent research.

The third stage is the pre-trained model (PLM) stage²².Unlike the aforementioned Markov chain model,PLM utilizes recurrent neural networks to capture word interactions within sentences²⁵, ²⁸.Subsequently, Google’s proposed Transformer architecture introduced self-attention mechanisms andpositional coding, which greatly improved the model’s parallel processing capabilities, enabling itto quickly learn a large amount of generalized knowledge and efficiently capture the logicalrelationships between words⁹⁰.This innovation has given rise to pre-trained language models such asBERT and GPT-2. By fine-tuning these pre-trained models on specific tasks, they have achievedsignificant performance gains in almost all downstream tasks of natural language processing (NLP).

The fourth stage is the Large Language Model (LLM) stage. In the current stage, the number ofparameters of the language model breaks through the level of billions, tens of billions and evenhundreds of billions. As the model parameters reach new orders of magnitude, the model experiences’emergence’ phenomenon, and the model performance shows a leapfrog improvement, requiring only asmall amount of learning to improve the model performance on a specific task⁴. For example, GPT-3 isable to solve specific tasks excellently with simple contextual learning, while GPT-2 performsrelatively poorly, highlighting the important impact of parameter size on model performance. Alongwith the release of ChatGPT, large-scale language models have become the hottest research in AI, withgreat achievements in areas such as medicine, finance, and autonomous driving⁹⁴. In 2023, the updateiteration length of large models reaches the unit of days, and their capabilities in multimodaldomains such as text, speech, image, and video have been rapidly improved and enriched with features,and they have achieved performance beyond the human average in several tasks¹⁴.

It can be seen that with the development of the field of Natural Language Processing, especially therecent emergence of Large Language Models, computers are equipped with the first condition for therealization of LLM-based agents: advanced comprehension capabilities, including in-depthunderstanding of written language, speech, images, and video, as well as the ability to accuratelyinterpret human intentions and generate autonomous feedback. This builds a “brain” for LLM-basedagents, and the next step is to focus on how to equip the brain with the ability to act.

2.2.2 LLM-based Agents

Agent belongs to the field of Artificial Intelligence, which aims to construct a human-likeintelligent entity capable of perceiving the environment in real time, making decisions and takingresponsive actions⁷³.The main difference between LLM-based agents and expert systems is reflected intheir autonomy, LLM-based agents are able to perceive the environment in real time, rely on theknowledge base and chain of thought, generate decisions in line with human expectations, and actaccordingly⁹⁷.Humans have always been committed to building LLM-based agents with an intelligencelevel comparable to their own and with the ability to act, and the development of LLM-based agentscan be divided into four main stages:

The first stage is the symbolic agents stage. In this early stage, researchers worked to createentities capable of making decisions⁹⁵.The initial strategy was to formulate a large number of rules,input judgment conditions and feedback into the computer, and make it respond according to theserules. During this period, expert systems were representative of symbolic LLM-based agents. Theadvantage of symbolic LLM-based agents is their excellent interpretability, which clearly revealsthe computer’s thought processes⁷⁵, ⁵⁷.However, their limitations are equally significant: the limitednature of the rule input makes it difficult to handle complex real-world inputs, and as the numberof rules increases, symbolic LLM-based agents are limited in their speed of response and are unableto react quickly to inputs⁴¹, ⁷⁴.

The second stage is the Reactive agents stage. Unlike symbolic agents, LLM-based agents are centeredon their immediate perception and response to changes in the environment¹⁰, ⁵⁴, ⁶³. Unlike symbolic LLM-basedagents, which focus on symbolic manipulation and complex logical reasoning, LLM-based agents focus onestablishing a direct mapping between inputs and environmental stimuli as well as between outputs andbehavioral responses⁷⁷, ⁹.The goal is to achieve accurate and rapid responses with minimal computational resources.

The third stage is the Reinforcement learning-based agents stage. Thanks to increased computationalpower, LLM-based agents are able to handle more complex tasks through interaction with theenvironment⁵⁹.The reinforcement learning framework emphasizes that LLM-based agents must learn to makedecisions to maximize the cumulative rewards under the constraints of the given tasks and rules whileinteracting with the environment, so as to ensure that the choices made at each step are the optimalsolutions in the current context⁸⁶.Initially, reinforcement learning LLM-based agents relied on basictechniques such as policy search and value function optimization, typically Q-learning⁹² and SARSA⁹³.Subsequently, with the advancement of deep learning techniques, deep reinforcement learning hasemerged, which combines the strengths of both and allows the agents to process high-dimensionalinputs and to learn higher-level policies, as shown by AlphaGo⁷⁹ and DQN⁶⁰. The main advantage of thisapproach is the ability of LLM-based agents to autonomously adapt to unknown environments withouthuman intervention. Nevertheless, reinforcement learning still faces challenges of long trainingcycles, inefficient sample utilization, and application in unstable environments.

The fourth stage is the Large Language Model-based agents (LLM-based agents) stage. As mentionedearlier, Large Language Models have demonstrated significant performance and can be utilized as thebrain of LLM-based agents⁴³.LLM-based agents extend perception and action capabilities throughstrategies such as multimodal perception¹⁰³, ³⁷and tool invocation⁴⁴, ³⁴, enhance planning and reasoningcapabilities through techniques such as thought chaining and problem decomposition³⁵, ¹⁰², and gain theability to interact with the environment by learning from and responding to feedback⁸⁴, ³⁶. LLM-basedagents acquire generalization capabilities by learning a large training set, thus enabling freeswitching between different tasks, and have been applied to various real-world tasks in areas suchas financial services, smart homes, education and training, healthcare, and intelligent customerservice.

It can be seen that LLM-based agents based on Large Language Model can already realize autonomousperception, planning, decision-making and action through natural language interaction. Although thereis still a long way to go before the realization of general artificial intelligence, the currentresearch has made computers initially equipped with two conditions for the realization of LLMagents-based agents, which not only have advanced comprehension capabilities, but also have theability to effectively invoke tools or devices to perceive the physical environment through sensorsand independently mobilize the appropriate tools in order to achieve self-regulation of the real world.

3 Methodology for Constructing Bridge Operation and maintenance LLM-based Agents

The integration of Large Language Models and agents is rapidly evolving, but in the O&M management ofbridge structures, complex bridge engineering knowledge and specialized O&M skills are required. Theacquisition and accurate construction of specialized knowledge is critical for the performance ofLLM-based agents in downstream tasks. Our goal is to embed specialized knowledge into intelligentagents to support automation and LLM-based agents for bridge O&M management. In this section, wedescribe the methodology for building LLM-based agents ontologies in specialized domains.

3.1 Knowledge Source

In order to realize the specialized functions of LLM-based agents, this paper suggests buildingspecialized domain intelligences using a combination of distributed knowledge, structured knowledge,and multi-round conversation data.

Distributed Knowledge

Distributed knowledge is i.e., natural linguistic expertise.Although current LLM-based agents showexcellent comprehension and generation capabilities in general-purpose domains, mainstream LargeLanguage Models lack specialized domain knowledge in the pre-training dataset, thus making itdifficult to meet the demands of practical engineering applications. China’s Bridge TechnicalCondition Evaluation Standard divides bridge inspections into daily inspections, frequentinspections, periodic inspections, and special inspections, and after decades of constructionin China’s bridge engineering field, tens of thousands of inspection reports have been accumulated,which can be used as a key source of data for LLM-based agents’ specialization. Meanwhile, textssuch as academic papers, specialized books and industry specifications in the field of bridges arealso the main sources of data. However, distributed knowledge alone does notachieve good results because the complexity and depth of the bridge O&M domain require LLM-basedagents to have structured knowledge to address their shortcomings in the depth of knowledgerepresentation. Therefore, in the bridge domain, LLM-based agents need to have both high-qualitydistributed and structured knowledge to achieve better performance.

Structured Knowledge

A typical expression of structured knowledge is the knowledge graph, a concept introduced by⁸¹, which is essentially a structured way of organizing knowledge. Knowledge graph can beregarded as a kind of knowledge relationship graph, consisting of nodes and edges, where nodesrepresent entities or concepts and edges represent attributes or relationships. The applicationof knowledge graph can effectively solve the problem of the depth of knowledge expression ofintelligences in the field of bridge O&M. Distributed knowledge by itself is not sufficient foragents to understand the logical relationships in bridge engineering and structural O&M. It isrecommended to construct a multilevel O&M knowledge base covering system-structure-components ofbridges, perception-evaluation-decision making in O&M, principle-expression-treatment measures ofbridge diseases, and preprocessing-threshold setting-abnormal value handling of bridge data in orderto enhance the accuracy and professionalism of the agents. However, even with the combination ofdistributed and structured knowledge, agents still face challenges in realizing the cyclic dialogdataset of inquiry/command-feedback/action.

Multi-round Dialog Dataset

Multi-round conversation datasets involve not only linguistic conversations, but alignment betweenLLM-based agents and human commands in a broad sense. At the Large Language Model level, in order tounderstand human intentions, a multi-round Q-A (query-feedback) dataset, i.e., a large amount ofquery and feedback data, is required⁸⁰, ⁹⁹, ¹.In this process, the design of prompt templates (prompt wordengineering) is crucial, which needs to guide the model to output the desired results withoutupdating the weights of the Large Language Model according to the actual needs of the specializeddomain. At the level of LLM-based agents, in order to realize the automatic maintenance of bridgestructures, multiple rounds of I-A (instruction-action) datasets, i.e., a large number of instructionand action datasets, are required²³.Currently, a variety of intelligent algorithms and advanceddevices have been developed in the field of bridge operation and maintenance, and the multi-roundI-A dataset can help LLM-based agents invoke these tools more effectively to accomplish theautomatic maintenance of bridges. It is worth noting that each tool invocation cannot be supportedby a large number of algorithms and hardware development techniques.

3.3 Qualitative Result Evaluation

The previous section of this paper introduced the methods for preparing and constructing the datarequired for LLM-based agents. Based on this, this section will focus on the assessment frameworkand evaluation criteria for LLM-based agents.

Along with the introduction of large-scale language models, several widely adopted evaluationstandards have been established in the industry, including MMLU³²,BIG-bench⁸³, and HELM⁷, as well as aseries of human benchmark tests and evaluation tests focusing on model-specific capabilities. Inaddition, there are evaluation benchmarks focusing on model-specific capabilities, such as theTyDiQA¹⁸benchmark focusing on knowledge application and the MGSM⁶⁸ benchmark focusing on mathematical reasoning. In order to conduct an effective assessment, the selection of benchmarks should be based on the specific objectives of the assessment.

While assessment benchmarks, as previously described, have been widely adopted in evaluating thecomprehension and production capabilities of large language models, they fail to adequately assessthe models’ ability to plan and make decisions in complex environments, as well as the ability ofLLM-based agents to act. As LLM-based agents technology advances, there is increasing academicinterest in assessing the responsiveness of LLM-based agents in complex environments⁵².

Currently, while there are no widely accepted benchmarks for the assessment of LLM-based agents,researchers have made significant progress in proposing candidate benchmarks that may become futureassessment standards. For example, ToolBench²⁹is a fine-tuned dataset for evaluating intelligences’ invocation of single- and multi-tool commands;TE³ evaluates the multifaceted ability of languagemodels to simulate human behavior; MetaTool³⁸aims to assess whether Large Language Models consciouslyinvoke tools and select the appropriate tool for problem solving; andLLM-Co² focuses on LLM-basedagents’ ability to infer partners’ intentions in games, engage in reasoning and the ability tocooperate over time. LLM-based agents are usually assessed on a task-specific basis and with acertain degree of ambiguity. As LLM-based agents, especially in the field of multi-intelligentcollaboration, methods for effectively tracking and evaluating the properties of intelligences willbecome increasingly important.

4 Potential Development Directions of LLM-based Agents in Bridge Operation and Maintenance

Automation and intelligence of bridge O&M is the key to improve O&M efficiency and enhance safetyand reliability. Through real-time monitoring, intelligent analysis and predictive maintenance, itrealizes optimal allocation of resources, reduces O&M costs, and enhances the public’s trust inbridge safety, which is an important development direction for bridge management in the future. Inthe field of bridge O&M, the integration of LLM-based agents is expected to reduce labor costs,reduce subjective judgments, improve management efficiency, and enable rapid response to emergencysituations. This section describes some of the research elements that need to be focused on forbridge O&M intelligence.

4.1 Large Language Model for Bridge Operations and Maintenance Domain

Currently, Large Language Models in vertical fields are booming. In China, a number of industrieshave launched customized large language models, such as “Hua Tuo” in the medical and healthindustry, “Xuan Yuan” in the financial field, “Han Fei” in the legal industry, and “Red Rabbit” in the field of intelligent customer service and marketing. Through deep learning and natural language processing technologies, these models accurately matchthe needs of various industries, providing a full range of intelligent solutions.

As an important transportation infrastructure, the efficiency of operation and maintenance managementof bridges is directly related to the safety, smoothness and efficiency of the transportationsystem. The construction of Large Language Model in the field of bridge O&M and its use as thecore hub of bridge intelligent management system can realize automated monitoring system andintelligent data analysis. The development and application of Bridge O&M Large Language Model isof great theoretical and practical significance to promote the development of bridge O&M managementin the direction of intelligence and efficiency.

4.2 Multimodal Knowledge Graph for Bridge Operations and Maintenance Domain

Building a multimodal knowledge graph in bridge operation and maintenance is a research anddevelopment effort aimed at integrating knowledge and data related to bridge operation andmaintenance, and enhancing the intelligence of bridge management through multimodal informationprocessing technology. Large Language Model demonstrates excellent multimodal processingcapabilities, especially in image and video processing⁴⁶, ⁵¹.However, in the field of bridge operationand maintenance, the processing capability of existing generalized Large Language Models for imagesis still at a low level. For example, in bridge disease picture recognition, even the best modelscan only recognize bridge cracks with unsatisfactory accuracy. There are a wide variety of bridgediseases, including cracked concrete, spalled and exposed reinforcement, weathering, honeycomb, andwater seepage in reinforced concrete bridges, as well as fracture, corrosion, coating defects, andsling relaxation in steel bridges, which cannot be effectively recognized by existing models.

To address this problem, combining the multimodal processing capability of Large Language Model withthe rich data in the field of bridge inspection, constructing a multimodal knowledge graph for bridgeO&M, establishing a hierarchical knowledge system, systematically classifying bridge diseases,components, parts, and structures, and realizing real-time processing of pictures and videoscollected on site will be a promising research direction. The construction of multimodal knowledgegraph in the field of bridge operation and maintenance can not only improve the intelligent levelof bridge operation and maintenance management, but also provide strong knowledge support for thewhole life cycle management of bridges.

4.3 Automatic Generation of Bridge Reports and Assisted Decision-Making

Currently, most knowledge management Q&A systems in the field of bridge maintenance are based onknowledge graphs¹⁰⁰, ⁵⁰, ⁶⁷.Knowledge graph-based Q&A systems need to extract structured knowledge from bridgedesign, construction, operation, management, inspection, and other data, a process that istime-consuming and labor-intensive, and difficult to handle unstructured or semi-structured data¹⁶.In addition, structured knowledge expressions may lead to error transmission and accumulation, andthe accuracy of knowledge extraction decreases dramatically when data complexity increases.

4.4 Tools Calling LLM-based Agents

In the field of bridge operation and maintenance, the introduction of Large Language Model-basedinteraction systems can gradually reshape its traditional paradigm. The Large Language Model-basedQ&A system adopts an unsupervised language model learning approach to map the knowledge to acontinuous numerical space, effectively overcoming a series of problems of traditional knowledgeextraction methods⁸⁵, ⁷⁸.With its excellent natural language understanding and generation capabilities,the system not only greatly improves the intelligence of information retrieval and knowledgeservices, but also promotes the automation and intelligence of operation and maintenance processes.O&M personnel can easily query all kinds of O&M records and reports through natural language, whileLarge Language Model can quickly provide precise information to assist decision-making, andautomatically mine potential problems from massive data to enhance the predictability of O&M.Meanwhile, the automated report generation function significantly reduces the time and errors ofmanual report writing and improves work efficiency.

Traditional bridge operation and maintenance management often relies on manual inspection and regulartesting, which is not only inefficient, but also difficult to comprehensively and accurately graspthe actual condition of the bridge. In recent years, with the improvement of bridge inspectionefficiency and quality needs, the development of intelligent inspection equipment and technologyhas gradually matured, covering drones, rope-climbing robots, underwater robots, sonar detectiondevices, as well as image acquisition technology, laser point-cloud scanning technology, holographicphotography technology and so on. The future research direction is bound to realize the automationand unmanned operation and maintenance management of bridges, and the application of LLM-based agentsbased on Large Language Model is expected to realize this goal.

Firstly, LLM-based agents realize accurate real-time monitoring of bridge structural health andenvironmental parameters by seamlessly integrating and scheduling a variety of monitoring systemsand sensors, utilizing Large Language Models to quickly analyze the data and discover potentialanomalies in a timely manner, ensuring the accuracy of monitoring. Secondly, LLM-based agents areexpected to realize automatic scheduling of advanced equipment such as drones and intelligent robotsto perform complex tasks such as high-altitude inspection and internal maintenance, realizingautomation and unmanned maintenance operations. This not only improves maintenance efficiency, butalso reduces personnel safety risks. These LLM-based agents maintain real-time communication with theintelligent body, ensuring accurate execution and efficient management of maintenance work.

5 Future Trends and Challenges

6 Conclusion

In the course of this study, we have successfully constructed a LLM-based agents, aiming to breakthrough the development bottleneck in the field of bridge operation and maintenance and to promotethe intelligent transformation of this industry. By comprehensively analyzing the current developmentstatus of the bridge O&M field and the advancement of LLM-based agents, we conclude thatLLM-based agents have significant advantages in understanding, generating, planning,decision-making, and action, and are able to effectively respond to the challenges facing the bridgeO&M field. The framework of LLM-based agents proposed in this study covers several key aspects suchas data sources, knowledge ontology construction and model evaluation, which provides strong supportfor the intelligent development of the bridge O&M domain. Meanwhile, we also explore the applicationprospect of this LLM-based agents in the bridge O&M field, and believe that it has a broad developmentspace. However, the application of LLM-based agents to the field of bridge operation andmaintenance still faces problems such as professional barriers and engineering ethics, which areimportant directions that need to be focused on and solved in future research. Overall, this studyprovides new ideas and methods for the development of intelligence in the field of bridge operationand maintenance, which is expected to promote the progress of the whole industry.

Acknowledgments

We thank the National Key Research and Development Program of China (2022YFC3801100) for financial support.

References

• Abbasiantaeb\BOthers. \APACyear2024\APACinsertmetastarabbasiantaeb2024let{APACrefauthors}Abbasiantaeb, Z., Yuan, Y., Kanoulas, E.\BCBL\BBA Aliannejadi, M.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleLet the llms talk: Simulating human-to-human conversational qa via zero-shot llm-to-llm interactions Let the llms talk: Simulating human-to-human conversational qa via zero-shot llm-to-llm interactions.\BBCQ\BIn \APACrefbtitleProceedings of the 17th ACM International Conference on Web Search and Data Mining Proceedings of the 17th acm international conference on web search and data mining(\BPGS8–17).\PrintBackRefs\CurrentBib
• Agashe\BOthers. \APACyear2023\APACinsertmetastaragashe2023evaluating{APACrefauthors}Agashe, S., Fan, Y.\BCBL\BBA Wang, X\BPBIE.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleEvaluating multi-agent coordination abilities in large language models Evaluating multi-agent coordination abilities in large language models.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2310.03903.\PrintBackRefs\CurrentBib
• Aher\BOthers. \APACyear2023\APACinsertmetastaraher2023using{APACrefauthors}Aher, G\BPBIV., Arriaga, R\BPBII.\BCBL\BBA Kalai, A\BPBIT.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleUsing large language models to simulate multiple humans and replicate human subject studies Using large language models to simulate multiple humans and replicate human subject studies.\BBCQ\BIn \APACrefbtitleInternational Conference on Machine Learning International conference on machine learning(\BPGS337–371).\PrintBackRefs\CurrentBib
• Beltagy\BOthers. \APACyear2019\APACinsertmetastarbeltagy2019scibert{APACrefauthors}Beltagy, I., Lo, K.\BCBL\BBA Cohan, A.\APACrefYearMonthDay2019.\BBOQ\APACrefatitleSciBERT: A pretrained language model for scientific text Scibert: A pretrained language model for scientific text.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:1903.10676.\PrintBackRefs\CurrentBib
• Bengio\BOthers. \APACyear2000\APACinsertmetastarbengio2000neural{APACrefauthors}Bengio, Y., Ducharme, R.\BCBL\BBA Vincent, P.\APACrefYearMonthDay2000.\BBOQ\APACrefatitleA neural probabilistic language model A neural probabilistic language model.\BBCQ\APACjournalVolNumPagesAdvances in neural information processing systems13.\PrintBackRefs\CurrentBib
• Bocchini\BBA Frangopol \APACyear2011\APACinsertmetastarbocchini2011probabilistic{APACrefauthors}Bocchini, P.\BCBT\BBA Frangopol, D\BPBIM.\APACrefYearMonthDay2011.\BBOQ\APACrefatitleA probabilistic computational framework for bridge network optimal maintenance scheduling A probabilistic computational framework for bridge network optimal maintenance scheduling.\BBCQ\APACjournalVolNumPagesReliability Engineering & System Safety962332–349.\PrintBackRefs\CurrentBib
• Bommasani\BOthers. \APACyear2023\APACinsertmetastarbommasani2023holistic{APACrefauthors}Bommasani, R., Liang, P.\BCBL\BBA Lee, T.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleHolistic evaluation of language models Holistic evaluation of language models.\BBCQ\APACjournalVolNumPagesAnnals of the New York Academy of Sciences15251140–146.\PrintBackRefs\CurrentBib
• Brockopp \APACyear1983\APACinsertmetastarbrockopp1983nlp{APACrefauthors}Brockopp, D\BPBIY.\APACrefYearMonthDay1983.\BBOQ\APACrefatitleWhat is NLP? What is nlp?\BBCQ\APACjournalVolNumPagesThe American Journal of Nursing8371012–1014.\PrintBackRefs\CurrentBib
• R.Brooks \APACyear1986\APACinsertmetastarbrooks1986robust{APACrefauthors}Brooks, R.\APACrefYearMonthDay1986.\BBOQ\APACrefatitleA robust layered control system for a mobile robot A robust layered control system for a mobile robot.\BBCQ\APACjournalVolNumPagesIEEE journal on robotics and automation2114–23.\PrintBackRefs\CurrentBib
• R\BPBIA.Brooks \APACyear1991\APACinsertmetastarbrooks1991intelligence{APACrefauthors}Brooks, R\BPBIA.\APACrefYearMonthDay1991.\BBOQ\APACrefatitleIntelligence without representation Intelligence without representation.\BBCQ\APACjournalVolNumPagesArtificial intelligence471-3139–159.\PrintBackRefs\CurrentBib
• P\BPBIF.Brown\BOthers. \APACyear1990\APACinsertmetastarbrown1990statistical{APACrefauthors}Brown, P\BPBIF., co*cke, J., DellaPietra, S\BPBIA., DellaPietra, V\BPBIJ., Jelinek, F., Lafferty, J.\BDBLRoossin, P\BPBIS.\APACrefYearMonthDay1990.\BBOQ\APACrefatitleA statistical approach to machine translation A statistical approach to machine translation.\BBCQ\APACjournalVolNumPagesComputational linguistics16279–85.\PrintBackRefs\CurrentBib
• T.Brown\BOthers. \APACyear2020\APACinsertmetastarbrown2020language{APACrefauthors}Brown, T., Mann, B., Ryder, N., Subbiah, M., Kaplan, J\BPBID., Dhariwal, P.\BDBLothers\APACrefYearMonthDay2020.\BBOQ\APACrefatitleLanguage models are few-shot learners Language models are few-shot learners.\BBCQ\APACjournalVolNumPagesAdvances in neural information processing systems331877–1901.\PrintBackRefs\CurrentBib
• Catbas\BBA Aktan \APACyear2002\APACinsertmetastarcatbas2002condition{APACrefauthors}Catbas, F\BPBIN.\BCBT\BBA Aktan, A\BPBIE.\APACrefYearMonthDay2002.\BBOQ\APACrefatitleCondition and damage assessment: issues and some promising indices Condition and damage assessment: issues and some promising indices.\BBCQ\APACjournalVolNumPagesJournal of Structural Engineering12881026–1036.\PrintBackRefs\CurrentBib
• Chang\BOthers. \APACyear2024\APACinsertmetastarchang2024survey{APACrefauthors}Chang, Y., Wang, X., Wang, J., Wu, Y., Yang, L., Zhu, K.\BDBLothers\APACrefYearMonthDay2024.\BBOQ\APACrefatitleA survey on evaluation of large language models A survey on evaluation of large language models.\BBCQ\APACjournalVolNumPagesACM Transactions on Intelligent Systems and Technology1531–45.\PrintBackRefs\CurrentBib
• S\BPBIF.Chen\BBA Goodman \APACyear1999\APACinsertmetastarchen1999empirical{APACrefauthors}Chen, S\BPBIF.\BCBT\BBA Goodman, J.\APACrefYearMonthDay1999.\BBOQ\APACrefatitleAn empirical study of smoothing techniques for language modeling An empirical study of smoothing techniques for language modeling.\BBCQ\APACjournalVolNumPagesComputer Speech & Language134359–394.\PrintBackRefs\CurrentBib
• X.Chen\BOthers. \APACyear2020\APACinsertmetastarchen2020review{APACrefauthors}Chen, X., Jia, S.\BCBL\BBA Xiang, Y.\APACrefYearMonthDay2020.\BBOQ\APACrefatitleA review: Knowledge reasoning over knowledge graph A review: Knowledge reasoning over knowledge graph.\BBCQ\APACjournalVolNumPagesExpert systems with applications141112948.\PrintBackRefs\CurrentBib
• Church \APACyear2017\APACinsertmetastarchurch2017word2vec{APACrefauthors}Church, K\BPBIW.\APACrefYearMonthDay2017.\BBOQ\APACrefatitleWord2Vec Word2vec.\BBCQ\APACjournalVolNumPagesNatural Language Engineering231155–162.\PrintBackRefs\CurrentBib
• Clark\BOthers. \APACyear2020\APACinsertmetastarclark2020tydi{APACrefauthors}Clark, J\BPBIH., Choi, E., Collins, M., Garrette, D., Kwiatkowski, T., Nikolaev, V.\BCBL\BBA Palomaki, J.\APACrefYearMonthDay2020.\BBOQ\APACrefatitleTydi qa: A benchmark for information-seeking question answering in ty pologically di verse languages Tydi qa: A benchmark for information-seeking question answering in ty pologically di verse languages.\BBCQ\APACjournalVolNumPagesTransactions of the Association for Computational Linguistics8454–470.\PrintBackRefs\CurrentBib
• Collobert\BBA Weston \APACyear2008\APACinsertmetastarcollobert2008unified{APACrefauthors}Collobert, R.\BCBT\BBA Weston, J.\APACrefYearMonthDay2008.\BBOQ\APACrefatitleA unified architecture for natural language processing: Deep neural networks with multitask learning A unified architecture for natural language processing: Deep neural networks with multitask learning.\BBCQ\BIn \APACrefbtitleProceedings of the 25th international conference on Machine learning Proceedings of the 25th international conference on machine learning(\BPGS160–167).\PrintBackRefs\CurrentBib
• Ding\BOthers. \APACyear2023\APACinsertmetastarding2023parameter{APACrefauthors}Ding, N., Qin, Y., Yang, G., Wei, F., Yang, Z., Su, Y.\BDBLothers\APACrefYearMonthDay2023.\BBOQ\APACrefatitleParameter-efficient fine-tuning of large-scale pre-trained language models Parameter-efficient fine-tuning of large-scale pre-trained language models.\BBCQ\APACjournalVolNumPagesNature Machine Intelligence53220–235.\PrintBackRefs\CurrentBib
• Dong \APACyear2024\APACinsertmetastardong2024multi{APACrefauthors}Dong, Y.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleThe Multi-agent System based on LLM for Online Discussions The multi-agent system based on llm for online discussions.\BBCQ\BIn \APACrefbtitleProceedings of the 23rd International Conference on Autonomous Agents and Multiagent Systems Proceedings of the 23rd international conference on autonomous agents and multiagent systems(\BPGS2731–2733).\PrintBackRefs\CurrentBib
• Erhan\BOthers. \APACyear2010\APACinsertmetastarerhan2010does{APACrefauthors}Erhan, D., Courville, A., Bengio, Y.\BCBL\BBA Vincent, P.\APACrefYearMonthDay2010.\BBOQ\APACrefatitleWhy does unsupervised pre-training help deep learning? Why does unsupervised pre-training help deep learning?\BBCQ\BIn \APACrefbtitleProceedings of the thirteenth international conference on artificial intelligence and statistics Proceedings of the thirteenth international conference on artificial intelligence and statistics(\BPGS201–208).\PrintBackRefs\CurrentBib
• Esteban-Lozano\BOthers. \APACyear2024\APACinsertmetastaresteban2024using{APACrefauthors}Esteban-Lozano, I., Castro-González, Á.\BCBL\BBA Martínez, P.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleUsing a LLM-Based Conversational Agent in the Social Robot Mini Using a llm-based conversational agent in the social robot mini.\BBCQ\BIn \APACrefbtitleInternational Conference on Human-Computer Interaction International conference on human-computer interaction(\BPGS15–26).\PrintBackRefs\CurrentBib
• Fujita\BOthers. \APACyear2024\APACinsertmetastarfujita2024llm{APACrefauthors}Fujita, M., Onaga, T.\BCBL\BBA Kano, Y.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleLLM Tuning and Interpretable CoT: KIS Team in COLIEE 2024 Llm tuning and interpretable cot: Kis team in coliee 2024.\BBCQ\BIn \APACrefbtitleJSAI International Symposium on Artificial Intelligence Jsai international symposium on artificial intelligence(\BPGS140–155).\PrintBackRefs\CurrentBib
• Gers\BBA Schmidhuber \APACyear2001\APACinsertmetastargers2001lstm{APACrefauthors}Gers, F\BPBIA.\BCBT\BBA Schmidhuber, E.\APACrefYearMonthDay2001.\BBOQ\APACrefatitleLSTM recurrent networks learn simple context-free and context-sensitive languages Lstm recurrent networks learn simple context-free and context-sensitive languages.\BBCQ\APACjournalVolNumPagesIEEE transactions on neural networks1261333–1340.\PrintBackRefs\CurrentBib
• Goodwin \APACyear1995\APACinsertmetastargoodwin1995formalizing{APACrefauthors}Goodwin, R.\APACrefYearMonthDay1995.\BBOQ\APACrefatitleFormalizing properties of agents Formalizing properties of agents.\BBCQ\APACjournalVolNumPagesJournal of Logic and Computation56763–781.\PrintBackRefs\CurrentBib
• Graves\BOthers. \APACyear2007\APACinsertmetastargraves2007unconstrained{APACrefauthors}Graves, A., Liwicki, M., Bunke, H., Schmidhuber, J.\BCBL\BBA Fernández, S.\APACrefYearMonthDay2007.\BBOQ\APACrefatitleUnconstrained on-line handwriting recognition with recurrent neural networks Unconstrained on-line handwriting recognition with recurrent neural networks.\BBCQ\APACjournalVolNumPagesAdvances in neural information processing systems20.\PrintBackRefs\CurrentBib
• Graves\BOthers. \APACyear2008\APACinsertmetastargraves2008novel{APACrefauthors}Graves, A., Liwicki, M., Fernández, S., Bertolami, R., Bunke, H.\BCBL\BBA Schmidhuber, J.\APACrefYearMonthDay2008.\BBOQ\APACrefatitleA novel connectionist system for unconstrained handwriting recognition A novel connectionist system for unconstrained handwriting recognition.\BBCQ\APACjournalVolNumPagesIEEE transactions on pattern analysis and machine intelligence315855–868.\PrintBackRefs\CurrentBib
• Guo\BOthers. \APACyear2024\APACinsertmetastarguo2024stabletoolbench{APACrefauthors}Guo, Z., Cheng, S., Wang, H., Liang, S., Qin, Y., Li, P.\BDBLLiu, Y.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleStableToolBench: Towards Stable Large-Scale Benchmarking on Tool Learning of Large Language Models Stabletoolbench: Towards stable large-scale benchmarking on tool learning of large language models.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2403.07714.\PrintBackRefs\CurrentBib
• Han\BOthers. \APACyear2021\APACinsertmetastarhan2021risk{APACrefauthors}Han, X., Yang, D\BPBIY.\BCBL\BBA Frangopol, D\BPBIM.\APACrefYearMonthDay2021.\BBOQ\APACrefatitleRisk-based life-cycle optimization of deteriorating steel bridges: Investigation on the use of novel corrosion resistant steel Risk-based life-cycle optimization of deteriorating steel bridges: Investigation on the use of novel corrosion resistant steel.\BBCQ\APACjournalVolNumPagesAdvances in Structural Engineering2481668–1686.\PrintBackRefs\CurrentBib
• He\BOthers. \APACyear2022\APACinsertmetastarhe2022integrated{APACrefauthors}He, Z., Li, W., Salehi, H., Zhang, H., Zhou, H.\BCBL\BBA Jiao, P.\APACrefYearMonthDay2022.\BBOQ\APACrefatitleIntegrated structural health monitoring in bridge engineering Integrated structural health monitoring in bridge engineering.\BBCQ\APACjournalVolNumPagesAutomation in construction136104168.\PrintBackRefs\CurrentBib
• Hendrycks\BOthers. \APACyear2020\APACinsertmetastarhendrycks2020measuring{APACrefauthors}Hendrycks, D., Burns, C., Basart, S., Zou, A., Mazeika, M., Song, D.\BCBL\BBA Steinhardt, J.\APACrefYearMonthDay2020.\BBOQ\APACrefatitleMeasuring massive multitask language understanding Measuring massive multitask language understanding.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2009.03300.\PrintBackRefs\CurrentBib
• Hiemstra \APACyear2001\APACinsertmetastarhiemstra2001using{APACrefauthors}Hiemstra, D.\APACrefYearMonthDay2001.\BBOQ\APACrefatitleUsing language models for information retrieval Using language models for information retrieval.\BBCQ\PrintBackRefs\CurrentBib
• Hong\BOthers. \APACyear2023\APACinsertmetastarhong2023metagpt{APACrefauthors}Hong, S., Zheng, X., Chen, J., Cheng, Y., Wang, J., Zhang, C.\BDBLothers\APACrefYearMonthDay2023.\BBOQ\APACrefatitleMetagpt: Meta programming for multi-agent collaborative framework Metagpt: Meta programming for multi-agent collaborative framework.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2308.00352.\PrintBackRefs\CurrentBib
• Hsieh\BOthers. \APACyear2023\APACinsertmetastarhsieh2023distilling{APACrefauthors}Hsieh, C\BHBIY., Li, C\BHBIL., Yeh, C\BHBIK., Nakhost, H., Fujii, Y., Ratner, A.\BDBLPfister, T.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleDistilling step-by-step! outperforming larger language models with less training data and smaller model sizes Distilling step-by-step! outperforming larger language models with less training data and smaller model sizes.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2305.02301.\PrintBackRefs\CurrentBib
• A\BPBIH.Huang\BOthers. \APACyear2023\APACinsertmetastarhuang2023finbert{APACrefauthors}Huang, A\BPBIH., Wang, H.\BCBL\BBA Yang, Y.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleFinBERT: A large language model for extracting information from financial text Finbert: A large language model for extracting information from financial text.\BBCQ\APACjournalVolNumPagesContemporary Accounting Research402806–841.\PrintBackRefs\CurrentBib
• S.Huang\BOthers. \APACyear2024\APACinsertmetastarhuang2024language{APACrefauthors}Huang, S., Dong, L., Wang, W., Hao, Y., Singhal, S., Ma, S.\BDBLothers\APACrefYearMonthDay2024.\BBOQ\APACrefatitleLanguage is not all you need: Aligning perception with language models Language is not all you need: Aligning perception with language models.\BBCQ\APACjournalVolNumPagesAdvances in Neural Information Processing Systems36.\PrintBackRefs\CurrentBib
• Y.Huang\BOthers. \APACyear2023\APACinsertmetastarhuang2023metatool{APACrefauthors}Huang, Y., Shi, J., Li, Y., Fan, C., Wu, S., Zhang, Q.\BDBLothers\APACrefYearMonthDay2023.\BBOQ\APACrefatitleMetatool benchmark for large language models: Deciding whether to use tools and which to use Metatool benchmark for large language models: Deciding whether to use tools and which to use.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2310.03128.\PrintBackRefs\CurrentBib
• Jensen \APACyear2020\APACinsertmetastarjensen2020innovative{APACrefauthors}Jensen, J\BPBIS.\APACrefYearMonthDay2020.\BBOQ\APACrefatitleInnovative and sustainable operation and maintenance of bridges Innovative and sustainable operation and maintenance of bridges.\BBCQ\APACjournalVolNumPagesStructure and Infrastructure Engineering16172–83.\PrintBackRefs\CurrentBib
• Jeong\BOthers. \APACyear2018\APACinsertmetastarjeong2018bridge{APACrefauthors}Jeong, Y., Kim, W., Lee, I.\BCBL\BBA Lee, J.\APACrefYearMonthDay2018.\BBOQ\APACrefatitleBridge inspection practices and bridge management programs in China, Japan, Korea, and US Bridge inspection practices and bridge management programs in china, japan, korea, and us.\BBCQ\APACjournalVolNumPagesJournal of Structural Integrity and Maintenance32126–135.\PrintBackRefs\CurrentBib
• Kaelbling\BOthers. \APACyear1987\APACinsertmetastarkaelbling1987architecture{APACrefauthors}Kaelbling, L\BPBIP.\BCBT\BOthersPeriod.\APACrefYearMonthDay1987.\BBOQ\APACrefatitleAn architecture for intelligent reactive systems An architecture for intelligent reactive systems.\BBCQ\APACjournalVolNumPagesReasoning about actions and plans395–410.\PrintBackRefs\CurrentBib
• Kansal \APACyear2024\APACinsertmetastarkansal2024langchain{APACrefauthors}Kansal, A.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleLangChain: Your Swiss Army Knife Langchain: Your swiss army knife.\BBCQ\BIn \APACrefbtitleBuilding Generative AI-Powered Apps: A Hands-on Guide for Developers Building generative ai-powered apps: A hands-on guide for developers(\BPGS17–40).\APACaddressPublisherSpringer.\PrintBackRefs\CurrentBib
• Kasneci\BOthers. \APACyear2023\APACinsertmetastarkasneci2023chatgpt{APACrefauthors}Kasneci, E., Seßler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F.\BDBLothers\APACrefYearMonthDay2023.\BBOQ\APACrefatitleChatGPT for good? On opportunities and challenges of large language models for education Chatgpt for good? on opportunities and challenges of large language models for education.\BBCQ\APACjournalVolNumPagesLearning and individual differences103102274.\PrintBackRefs\CurrentBib
• Kim\BOthers. \APACyear2023\APACinsertmetastarkim2023llm{APACrefauthors}Kim, S., Moon, S., Tabrizi, R., Lee, N., Mahoney, M\BPBIW., Keutzer, K.\BCBL\BBA Gholami, A.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleAn LLM compiler for parallel function calling An llm compiler for parallel function calling.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2312.04511.\PrintBackRefs\CurrentBib
• Kneser\BBA Ney \APACyear1995\APACinsertmetastarkneser1995improved{APACrefauthors}Kneser, R.\BCBT\BBA Ney, H.\APACrefYearMonthDay1995.\BBOQ\APACrefatitleImproved backing-off for m-gram language modeling Improved backing-off for m-gram language modeling.\BBCQ\BIn \APACrefbtitle1995 international conference on acoustics, speech, and signal processing 1995 international conference on acoustics, speech, and signal processing(\BVOL1, \BPGS181–184).\PrintBackRefs\CurrentBib
• Koh\BOthers. \APACyear2024\APACinsertmetastarkoh2024generating{APACrefauthors}Koh, J\BPBIY., Fried, D.\BCBL\BBA Salakhutdinov, R\BPBIR.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleGenerating images with multimodal language models Generating images with multimodal language models.\BBCQ\APACjournalVolNumPagesAdvances in Neural Information Processing Systems36.\PrintBackRefs\CurrentBib
• Lafferty\BBA Zhai \APACyear2001\APACinsertmetastarlafferty2001document{APACrefauthors}Lafferty, J.\BCBT\BBA Zhai, C.\APACrefYearMonthDay2001.\BBOQ\APACrefatitleDocument language models, query models, and risk minimization for information retrieval Document language models, query models, and risk minimization for information retrieval.\BBCQ\BIn \APACrefbtitleProceedings of the 24th annual international ACM SIGIR conference on Research and development in information retrieval Proceedings of the 24th annual international acm sigir conference on research and development in information retrieval(\BPGS111–119).\PrintBackRefs\CurrentBib
• Lewis\BOthers. \APACyear2020\APACinsertmetastarlewis2020retrieval{APACrefauthors}Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N.\BDBLothers\APACrefYearMonthDay2020.\BBOQ\APACrefatitleRetrieval-augmented generation for knowledge-intensive nlp tasks Retrieval-augmented generation for knowledge-intensive nlp tasks.\BBCQ\APACjournalVolNumPagesAdvances in Neural Information Processing Systems339459–9474.\PrintBackRefs\CurrentBib
• Lilleberg\BOthers. \APACyear2015\APACinsertmetastarlilleberg2015support{APACrefauthors}Lilleberg, J., Zhu, Y.\BCBL\BBA Zhang, Y.\APACrefYearMonthDay2015.\BBOQ\APACrefatitleSupport vector machines and word2vec for text classification with semantic features Support vector machines and word2vec for text classification with semantic features.\BBCQ\BIn \APACrefbtitle2015 IEEE 14th International Conference on Cognitive Informatics & Cognitive Computing (ICCI* CC) 2015 ieee 14th international conference on cognitive informatics & cognitive computing (icci* cc)(\BPGS136–140).\PrintBackRefs\CurrentBib
• J.Liu\BOthers. \APACyear2021\APACinsertmetastarliu2021knowledge{APACrefauthors}Liu, J., Schmid, F., Li, K.\BCBL\BBA Zheng, W.\APACrefYearMonthDay2021.\BBOQ\APACrefatitleA knowledge graph-based approach for exploring railway operational accidents A knowledge graph-based approach for exploring railway operational accidents.\BBCQ\APACjournalVolNumPagesReliability engineering & system safety207107352.\PrintBackRefs\CurrentBib
• Y.Liu\BOthers. \APACyear2023\APACinsertmetastarliu2023summary{APACrefauthors}Liu, Y., Han, T., Ma, S., Zhang, J., Yang, Y., Tian, J.\BDBLothers\APACrefYearMonthDay2023.\BBOQ\APACrefatitleSummary of chatgpt-related research and perspective towards the future of large language models Summary of chatgpt-related research and perspective towards the future of large language models.\BBCQ\APACjournalVolNumPagesMeta-Radiology100017.\PrintBackRefs\CurrentBib
• Y.Liu\BOthers. \APACyear2024\APACinsertmetastarliu2024understanding{APACrefauthors}Liu, Y., He, H., Han, T., Zhang, X., Liu, M., Tian, J.\BDBLothers\APACrefYearMonthDay2024.\BBOQ\APACrefatitleUnderstanding llms: A comprehensive overview from training to inference Understanding llms: A comprehensive overview from training to inference.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2401.02038.\PrintBackRefs\CurrentBib
• Ma\BBA Zhang \APACyear2015\APACinsertmetastarma2015using{APACrefauthors}Ma, L.\BCBT\BBA Zhang, Y.\APACrefYearMonthDay2015.\BBOQ\APACrefatitleUsing Word2Vec to process big text data Using word2vec to process big text data.\BBCQ\BIn \APACrefbtitle2015 IEEE International Conference on Big Data (Big Data) 2015 ieee international conference on big data (big data)(\BPGS2895–2897).\PrintBackRefs\CurrentBib
• Maes \APACyear1990\APACinsertmetastarmaes1990designing{APACrefauthors}Maes, P.\APACrefYear1990.\APACrefbtitleDesigning autonomous agents: Theory and practice from biology to engineering and back Designing autonomous agents: Theory and practice from biology to engineering and back.\APACaddressPublisherMIT press.\PrintBackRefs\CurrentBib
• MandićIvanković\BOthers. \APACyear2019\APACinsertmetastarmandic2019european{APACrefauthors}MandićIvanković, A., Strauss, A.\BCBL\BBA Sousa, H.\APACrefYearMonthDay2019.\BBOQ\APACrefatitleEuropean review of performance indicators towards sustainable road bridge management European review of performance indicators towards sustainable road bridge management.\BBCQ\BIn \APACrefbtitleProceedings of the Institution of Civil Engineers-Engineering Sustainability Proceedings of the institution of civil engineers-engineering sustainability(\BVOL173, \BPGS109–124).\PrintBackRefs\CurrentBib
• M.Bran\BOthers. \APACyear2024\APACinsertmetastarm2024augmenting{APACrefauthors}M.Bran, A., Cox, S., Schilter, O., Baldassari, C., White, A\BPBID.\BCBL\BBA Schwaller, P.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleAugmenting large language models with chemistry tools Augmenting large language models with chemistry tools.\BBCQ\APACjournalVolNumPagesNature Machine Intelligence1–11.\PrintBackRefs\CurrentBib
• McAllester\BBA Rosenblatt \APACyear1991\APACinsertmetastarmcallester1991systematic{APACrefauthors}McAllester, D.\BCBT\BBA Rosenblatt, D.\APACrefYearMonthDay1991.\BBOQ\APACrefatitleSystematic nonlinear planning Systematic nonlinear planning.\BBCQ\PrintBackRefs\CurrentBib
• Mikolov\BOthers. \APACyear2010\APACinsertmetastarmikolov2010recurrent{APACrefauthors}Mikolov, T., Karafiát, M., Burget, L., Cernockỳ, J.\BCBL\BBA Khudanpur, S.\APACrefYearMonthDay2010.\BBOQ\APACrefatitleRecurrent neural network based language model. Recurrent neural network based language model.\BBCQ\BIn \APACrefbtitleInterspeech Interspeech(\BVOL2, \BPGS1045–1048).\PrintBackRefs\CurrentBib
• Minsky \APACyear1961\APACinsertmetastarminsky1961steps{APACrefauthors}Minsky, M.\APACrefYearMonthDay1961.\BBOQ\APACrefatitleSteps toward artificial intelligence Steps toward artificial intelligence.\BBCQ\APACjournalVolNumPagesProceedings of the IRE4918–30.\PrintBackRefs\CurrentBib
• Mnih\BOthers. \APACyear2013\APACinsertmetastarmnih2013playing{APACrefauthors}Mnih, V., Kavukcuoglu, K., Silver, D., Graves, A., Antonoglou, I., Wierstra, D.\BCBL\BBA Riedmiller, M.\APACrefYearMonthDay2013.\BBOQ\APACrefatitlePlaying atari with deep reinforcement learning Playing atari with deep reinforcement learning.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:1312.5602.\PrintBackRefs\CurrentBib
• Morin\BBA Bengio \APACyear2005\APACinsertmetastarmorin2005hierarchical{APACrefauthors}Morin, F.\BCBT\BBA Bengio, Y.\APACrefYearMonthDay2005.\BBOQ\APACrefatitleHierarchical probabilistic neural network language model Hierarchical probabilistic neural network language model.\BBCQ\BIn \APACrefbtitleInternational workshop on artificial intelligence and statistics International workshop on artificial intelligence and statistics(\BPGS246–252).\PrintBackRefs\CurrentBib
• Nadkarni\BOthers. \APACyear2011\APACinsertmetastarnadkarni2011natural{APACrefauthors}Nadkarni, P\BPBIM., Ohno-Machado, L.\BCBL\BBA Chapman, W\BPBIW.\APACrefYearMonthDay2011.\BBOQ\APACrefatitleNatural language processing: an introduction Natural language processing: an introduction.\BBCQ\APACjournalVolNumPagesJournal of the American Medical Informatics Association185544–551.\PrintBackRefs\CurrentBib
• Nilsson \APACyear1992\APACinsertmetastarnilsson1992toward{APACrefauthors}Nilsson, N\BPBIJ.\APACrefYearMonthDay1992.\APACrefbtitleToward agent programs with circuit semantics Toward agent programs with circuit semantics\APACbVolEdTR\BTR.\PrintBackRefs\CurrentBib
• Nogueri Alonso \APACyear1950\APACinsertmetastarnoguer1950key{APACrefauthors}Nogueri Alonso, M.\APACrefYearMonthDay1950.\BBOQ\APACrefatitleKey Milestones in Natural Language Processing (NLP) 1950-2024 Key milestones in natural language processing (nlp) 1950-2024.\BBCQ\APACjournalVolNumPagesKey Milestones in Natural Language Processing (NLP)2024.\PrintBackRefs\CurrentBib
• Otieno\BOthers. \APACyear2024\APACinsertmetastarotieno2024accuracy{APACrefauthors}Otieno, D\BPBIO., Abri, F., Siami-Namini, S.\BCBL\BBA Namin, A\BPBIS.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleThe Accuracy of Domain Specific and Descriptive Analysis Generated by Large Language Models The accuracy of domain specific and descriptive analysis generated by large language models.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2405.19578.\PrintBackRefs\CurrentBib
• Ouyang\BOthers. \APACyear2022\APACinsertmetastarouyang2022training{APACrefauthors}Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P.\BDBLothers\APACrefYearMonthDay2022.\BBOQ\APACrefatitleTraining language models to follow instructions with human feedback Training language models to follow instructions with human feedback.\BBCQ\APACjournalVolNumPagesAdvances in neural information processing systems3527730–27744.\PrintBackRefs\CurrentBib
• Park\BOthers. \APACyear2021\APACinsertmetastarpark2021bridge{APACrefauthors}Park, J., Lee, J.\BCBL\BBA Sohn, K.\APACrefYearMonthDay2021.\BBOQ\APACrefatitleBridge to answer: Structure-aware graph interaction network for video question answering Bridge to answer: Structure-aware graph interaction network for video question answering.\BBCQ\BIn \APACrefbtitleProceedings of the IEEE/CVF conference on computer vision and pattern recognition Proceedings of the ieee/cvf conference on computer vision and pattern recognition(\BPGS15526–15535).\PrintBackRefs\CurrentBib
• Qian\BOthers. \APACyear2022\APACinsertmetastarqian2022limitations{APACrefauthors}Qian, J., Wang, H., Li, Z., Li, S.\BCBL\BBA Yan, X.\APACrefYearMonthDay2022.\BBOQ\APACrefatitleLimitations of language models in arithmetic and symbolic induction Limitations of language models in arithmetic and symbolic induction.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2208.05051.\PrintBackRefs\CurrentBib
• Radiya-Dixit\BBA Wang \APACyear2020\APACinsertmetastarradiya2020fine{APACrefauthors}Radiya-Dixit, E.\BCBT\BBA Wang, X.\APACrefYearMonthDay2020.\BBOQ\APACrefatitleHow fine can fine-tuning be? learning efficient language models How fine can fine-tuning be? learning efficient language models.\BBCQ\BIn \APACrefbtitleInternational Conference on Artificial Intelligence and Statistics International conference on artificial intelligence and statistics(\BPGS2435–2443).\PrintBackRefs\CurrentBib
• Rizzo\BBA Enshaeian \APACyear2021\APACinsertmetastarrizzo2021challenges{APACrefauthors}Rizzo, P.\BCBT\BBA Enshaeian, A.\APACrefYearMonthDay2021.\BBOQ\APACrefatitleChallenges in bridge health monitoring: A review Challenges in bridge health monitoring: A review.\BBCQ\APACjournalVolNumPagesSensors21134336.\PrintBackRefs\CurrentBib
• Rodriguez\BOthers. \APACyear1999\APACinsertmetastarrodriguez1999recurrent{APACrefauthors}Rodriguez, P., Wiles, J.\BCBL\BBA Elman, J\BPBIL.\APACrefYearMonthDay1999.\BBOQ\APACrefatitleA recurrent neural network that learns to count A recurrent neural network that learns to count.\BBCQ\APACjournalVolNumPagesConnection Science1115–40.\PrintBackRefs\CurrentBib
• Rosenfeld \APACyear2000\APACinsertmetastarrosenfeld2000two{APACrefauthors}Rosenfeld, R.\APACrefYearMonthDay2000.\BBOQ\APACrefatitleTwo decades of statistical language modeling: Where do we go from here? Two decades of statistical language modeling: Where do we go from here?\BBCQ\APACjournalVolNumPagesProceedings of the IEEE8881270–1278.\PrintBackRefs\CurrentBib
• Russell\BBA Norvig \APACyear2016\APACinsertmetastarrussell2016artificial{APACrefauthors}Russell, S\BPBIJ.\BCBT\BBA Norvig, P.\APACrefYear2016.\APACrefbtitleArtificial intelligence: a modern approach Artificial intelligence: a modern approach.\APACaddressPublisherPearson.\PrintBackRefs\CurrentBib
• Russell\BBA Wefald \APACyear1991\APACinsertmetastarrussell1991right{APACrefauthors}Russell, S\BPBIJ.\BCBT\BBA Wefald, E.\APACrefYear1991.\APACrefbtitleDo the right thing: studies in limited rationality Do the right thing: studies in limited rationality.\APACaddressPublisherMIT press.\PrintBackRefs\CurrentBib
• Sacerdoti \APACyear1975\APACinsertmetastarsacerdoti1975nonlinear{APACrefauthors}Sacerdoti, E\BPBID.\APACrefYearMonthDay1975.\BBOQ\APACrefatitleThe nonlinear nature of plans The nonlinear nature of plans.\BBCQ\PrintBackRefs\CurrentBib
• Sasmal\BBA Ramanjaneyulu \APACyear2008\APACinsertmetastarsasmal2008condition{APACrefauthors}Sasmal, S.\BCBT\BBA Ramanjaneyulu, K.\APACrefYearMonthDay2008.\BBOQ\APACrefatitleCondition evaluation of existing reinforced concrete bridges using fuzzy based analytic hierarchy approach Condition evaluation of existing reinforced concrete bridges using fuzzy based analytic hierarchy approach.\BBCQ\APACjournalVolNumPagesExpert Systems with Applications3531430–1443.\PrintBackRefs\CurrentBib
• Schoppers \APACyear1987\APACinsertmetastarschoppers1987universal{APACrefauthors}Schoppers, M.\APACrefYearMonthDay1987.\BBOQ\APACrefatitleUniversal Plans for Reactive Robots in Unpredictable Environments. Universal plans for reactive robots in unpredictable environments.\BBCQ\BIn \APACrefbtitleIJCAI Ijcai(\BVOL87, \BPGS1039–1046).\PrintBackRefs\CurrentBib
• Shao\BOthers. \APACyear2023\APACinsertmetastarshao2023prompting{APACrefauthors}Shao, Z., Yu, Z., Wang, M.\BCBL\BBA Yu, J.\APACrefYearMonthDay2023.\BBOQ\APACrefatitlePrompting large language models with answer heuristics for knowledge-based visual question answering Prompting large language models with answer heuristics for knowledge-based visual question answering.\BBCQ\BIn \APACrefbtitleProceedings of the IEEE/CVF Conference on computer vision and pattern recognition Proceedings of the ieee/cvf conference on computer vision and pattern recognition(\BPGS14974–14983).\PrintBackRefs\CurrentBib
• Silver\BOthers. \APACyear2016\APACinsertmetastarsilver2016mastering{APACrefauthors}Silver, D., Huang, A., Maddison, C\BPBIJ., Guez, A., Sifre, L., Van DenDriessche, G.\BDBLothers\APACrefYearMonthDay2016.\BBOQ\APACrefatitleMastering the game of Go with deep neural networks and tree search Mastering the game of go with deep neural networks and tree search.\BBCQ\APACjournalVolNumPagesnature5297587484–489.\PrintBackRefs\CurrentBib
• Singh\BOthers. \APACyear2019\APACinsertmetastarsingh2019strings{APACrefauthors}Singh, A\BPBIK., Mishra, A., Shekhar, S.\BCBL\BBA Chakraborty, A.\APACrefYearMonthDay2019.\BBOQ\APACrefatitleFrom strings to things: Knowledge-enabled vqa model that can read and reason From strings to things: Knowledge-enabled vqa model that can read and reason.\BBCQ\BIn \APACrefbtitleProceedings of the IEEE/CVF international conference on computer vision Proceedings of the ieee/cvf international conference on computer vision(\BPGS4602–4612).\PrintBackRefs\CurrentBib
• Singhal \APACyear2012\APACinsertmetastarSinghal2012Official{APACrefauthors}Singhal.\APACrefYearMonthDay2012.\BBOQ\APACrefatitleOfficial Google Blog: Introducing the Knowledge Graph: things, not strings Official google blog: Introducing the knowledge graph: things, not strings.\BBCQ\PrintBackRefs\CurrentBib
• Söderqvist\BBA Veijola \APACyear1998\APACinsertmetastarsoderqvist1998finnish{APACrefauthors}Söderqvist, M\BHBIK.\BCBT\BBA Veijola, M.\APACrefYearMonthDay1998.\BBOQ\APACrefatitleThe Finnish bridge management system The finnish bridge management system.\BBCQ\APACjournalVolNumPagesStructural engineering international84315–319.\PrintBackRefs\CurrentBib
• Srivastava\BOthers. \APACyear2022\APACinsertmetastarsrivastava2022beyond{APACrefauthors}Srivastava, A., Rastogi, A., Rao, A., Shoeb, A\BPBIA\BPBIM., Abid, A., Fisch, A.\BDBLothers\APACrefYearMonthDay2022.\BBOQ\APACrefatitleBeyond the imitation game: Quantifying and extrapolating the capabilities of language models Beyond the imitation game: Quantifying and extrapolating the capabilities of language models.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2206.04615.\PrintBackRefs\CurrentBib
• Stamper\BOthers. \APACyear2024\APACinsertmetastarstamper2024enhancing{APACrefauthors}Stamper, J., Xiao, R.\BCBL\BBA Hou, X.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleEnhancing LLM-Based Feedback: Insights from Intelligent Tutoring Systems and the Learning Sciences Enhancing llm-based feedback: Insights from intelligent tutoring systems and the learning sciences.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2405.04645.\PrintBackRefs\CurrentBib
• Su\BOthers. \APACyear2019\APACinsertmetastarsu2019generalizing{APACrefauthors}Su, D., Xu, Y., Winata, G\BPBII., Xu, P., Kim, H., Liu, Z.\BCBL\BBA Fung, P.\APACrefYearMonthDay2019.\BBOQ\APACrefatitleGeneralizing question answering system with pre-trained language model fine-tuning Generalizing question answering system with pre-trained language model fine-tuning.\BBCQ\BIn \APACrefbtitleProceedings of the 2nd workshop on machine reading for question answering Proceedings of the 2nd workshop on machine reading for question answering(\BPGS203–211).\PrintBackRefs\CurrentBib
• Sutton\BBA Barto \APACyear2018\APACinsertmetastarsutton2018reinforcement{APACrefauthors}Sutton, R\BPBIS.\BCBT\BBA Barto, A\BPBIG.\APACrefYear2018.\APACrefbtitleReinforcement learning: An introduction Reinforcement learning: An introduction.\APACaddressPublisherMIT press.\PrintBackRefs\CurrentBib
• Thirunavukarasu\BOthers. \APACyear2023\APACinsertmetastarthirunavukarasu2023large{APACrefauthors}Thirunavukarasu, A\BPBIJ., Ting, D\BPBIS\BPBIJ., Elangovan, K., Gutierrez, L., Tan, T\BPBIF.\BCBL\BBA Ting, D\BPBIS\BPBIW.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleLarge language models in medicine Large language models in medicine.\BBCQ\APACjournalVolNumPagesNature medicine2981930–1940.\PrintBackRefs\CurrentBib
• Thompson\BOthers. \APACyear1998\APACinsertmetastarthompson1998pontis{APACrefauthors}Thompson, P\BPBID., Small, E\BPBIP., Johnson, M.\BCBL\BBA Marshall, A\BPBIR.\APACrefYearMonthDay1998.\BBOQ\APACrefatitleThe Pontis bridge management system The pontis bridge management system.\BBCQ\APACjournalVolNumPagesStructural engineering international84303–308.\PrintBackRefs\CurrentBib
• Turian\BOthers. \APACyear2010\APACinsertmetastarturian2010word{APACrefauthors}Turian, J., Ratinov, L.\BCBL\BBA Bengio, Y.\APACrefYearMonthDay2010.\BBOQ\APACrefatitleWord representations: a simple and general method for semi-supervised learning Word representations: a simple and general method for semi-supervised learning.\BBCQ\BIn \APACrefbtitleProceedings of the 48th annual meeting of the association for computational linguistics Proceedings of the 48th annual meeting of the association for computational linguistics(\BPGS384–394).\PrintBackRefs\CurrentBib
• Vaswani\BOthers. \APACyear2017\APACinsertmetastarvaswani2017attention{APACrefauthors}Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A\BPBIN.\BDBLPolosukhin, I.\APACrefYearMonthDay2017.\BBOQ\APACrefatitleAttention is all you need Attention is all you need.\BBCQ\APACjournalVolNumPagesAdvances in neural information processing systems30.\PrintBackRefs\CurrentBib
• Wang\BOthers. \APACyear2024\APACinsertmetastarwang2024survey{APACrefauthors}Wang, L., Ma, C., Feng, X., Zhang, Z., Yang, H., Zhang, J.\BDBLothers\APACrefYearMonthDay2024.\BBOQ\APACrefatitleA survey on large language model based autonomous agents A survey on large language model based autonomous agents.\BBCQ\APACjournalVolNumPagesFrontiers of Computer Science186186345.\PrintBackRefs\CurrentBib
• C\BPBIJ.Watkins\BBA Dayan \APACyear1992\APACinsertmetastarwatkins1992q{APACrefauthors}Watkins, C\BPBIJ.\BCBT\BBA Dayan, P.\APACrefYearMonthDay1992.\BBOQ\APACrefatitleQ-learning Q-learning.\BBCQ\APACjournalVolNumPagesMachine learning8279–292.\PrintBackRefs\CurrentBib
• C\BPBIJ\BPBIC\BPBIH.Watkins \APACyear1989\APACinsertmetastarwatkins1989learning{APACrefauthors}Watkins, C\BPBIJ\BPBIC\BPBIH.\APACrefYearMonthDay1989.\BBOQ\APACrefatitleLearning from delayed rewards Learning from delayed rewards.\BBCQ\PrintBackRefs\CurrentBib
• Wei\BOthers. \APACyear2022\APACinsertmetastarwei2022chain{APACrefauthors}Wei, J., Wang, X., Schuurmans, D., Bosma, M., Xia, F., Chi, E.\BDBLothers\APACrefYearMonthDay2022.\BBOQ\APACrefatitleChain-of-thought prompting elicits reasoning in large language models Chain-of-thought prompting elicits reasoning in large language models.\BBCQ\APACjournalVolNumPagesAdvances in neural information processing systems3524824–24837.\PrintBackRefs\CurrentBib
• Wilkins \APACyear2014\APACinsertmetastarwilkins2014practical{APACrefauthors}Wilkins, D\BPBIE.\APACrefYear2014.\APACrefbtitlePractical planning: extending the classical AI planning paradigm Practical planning: extending the classical ai planning paradigm.\APACaddressPublisherElsevier.\PrintBackRefs\CurrentBib
• Woodridge\BBA Jennings \APACyear1995\APACinsertmetastarwoodridge1995intelligent{APACrefauthors}Woodridge, M.\BCBT\BBA Jennings, N.\APACrefYearMonthDay1995.\BBOQ\APACrefatitleIntelligent Agents: Theory and Practice The Knowledge Engineering Review Intelligent agents: Theory and practice the knowledge engineering review.\BBCQ\PrintBackRefs\CurrentBib
• Wooldridge\BBA Jennings \APACyear1995\APACinsertmetastarwooldridge1995intelligent{APACrefauthors}Wooldridge, M.\BCBT\BBA Jennings, N\BPBIR.\APACrefYearMonthDay1995.\BBOQ\APACrefatitleIntelligent agents: Theory and practice Intelligent agents: Theory and practice.\BBCQ\APACjournalVolNumPagesThe knowledge engineering review102115–152.\PrintBackRefs\CurrentBib
• Yang\BOthers. \APACyear2024\APACinsertmetastaryang2024harnessing{APACrefauthors}Yang, J., Jin, H., Tang, R., Han, X., Feng, Q., Jiang, H.\BDBLHu, X.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleHarnessing the power of llms in practice: A survey on chatgpt and beyond Harnessing the power of llms in practice: A survey on chatgpt and beyond.\BBCQ\APACjournalVolNumPagesACM Transactions on Knowledge Discovery from Data1861–32.\PrintBackRefs\CurrentBib
• Yi\BOthers. \APACyear2024\APACinsertmetastaryi2024survey{APACrefauthors}Yi, Z., Ouyang, J., Liu, Y., Liao, T., Xu, Z.\BCBL\BBA Shen, Y.\APACrefYearMonthDay2024.\BBOQ\APACrefatitleA Survey on Recent Advances in LLM-Based Multi-turn Dialogue Systems A survey on recent advances in llm-based multi-turn dialogue systems.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2402.18013.\PrintBackRefs\CurrentBib
• Zareian\BOthers. \APACyear2020\APACinsertmetastarzareian2020bridging{APACrefauthors}Zareian, A., Karaman, S.\BCBL\BBA Chang, S\BHBIF.\APACrefYearMonthDay2020.\BBOQ\APACrefatitleBridging knowledge graphs to generate scene graphs Bridging knowledge graphs to generate scene graphs.\BBCQ\BIn \APACrefbtitleComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXIII 16 Computer vision–eccv 2020: 16th european conference, glasgow, uk, august 23–28, 2020, proceedings, part xxiii 16(\BPGS606–623).\PrintBackRefs\CurrentBib
• Zhai\BOthers. \APACyear2008\APACinsertmetastarzhai2008statistical{APACrefauthors}Zhai, C.\BCBT\BOthersPeriod.\APACrefYearMonthDay2008.\BBOQ\APACrefatitleStatistical language models for information retrieval a critical review Statistical language models for information retrieval a critical review.\BBCQ\APACjournalVolNumPagesFoundations and Trends® in Information Retrieval23137–213.\PrintBackRefs\CurrentBib
• Zhang\BOthers. \APACyear2023\APACinsertmetastarzhang2023multimodal{APACrefauthors}Zhang, Z., Zhang, A., Li, M., Zhao, H., Karypis, G.\BCBL\BBA Smola, A.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleMultimodal chain-of-thought reasoning in language models Multimodal chain-of-thought reasoning in language models.\BBCQ\APACjournalVolNumPagesarXiv preprint arXiv:2302.00923.\PrintBackRefs\CurrentBib
• Zhao\BOthers. \APACyear2023\APACinsertmetastarzhao2023chat{APACrefauthors}Zhao, X., Li, M., Weber, C., Hafez, M\BPBIB.\BCBL\BBA Wermter, S.\APACrefYearMonthDay2023.\BBOQ\APACrefatitleChat with the environment: Interactive multimodal perception using large language models Chat with the environment: Interactive multimodal perception using large language models.\BBCQ\BIn \APACrefbtitle2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2023 ieee/rsj international conference on intelligent robots and systems (iros)(\BPGS3590–3596).\PrintBackRefs\CurrentBib