ALBERT with Knowledge Graph Encoder Utilizing Semantic Similarity for Commonsense Question Answering

Recently,pre-trained language representation models such as bidirec-tional encoder representations from transformers(BERT)have been performing well in commonsense question answering(CSQA).However,there is a problem that the models do not directly use explicit information of knowledge sources existing outside.To augment this,additional methods such as knowledge-aware graph network(KagNet)and multi-hop graph relation network(MHGRN)have been proposed.In this study,we propose to use the latest pre-trained language model a lite bidirectional encoder representations from transformers(ALBERT)with knowledge graph information extraction technique.We also propose to applying the novel method,schema graph expansion to recent language models.Then,we analyze the effect of applying knowledge graph-based knowledge extraction techniques to recent pre-trained language models and confirm that schema graph expansion is effective in some extent.Furthermore,we show that our proposed model can achieve better performance than existing KagNet and MHGRN models in CommonsenseQA dataset.

Commonsense Reasoning and Nonmonotonic Aggregation Connectives on Fuzzy Sets

Commonsense representation and manipulation based on fuzzy logic is a new research field which handles the incompleteness, error-tolerability (allow exceptions) and uncertainty associated with commonsense knowledge. In this paper, we introduce a pair of nonmonotonic aggregation connectives on fuzzy sets-soft intersection and soft union, in the light of Zadeh’s fuzzy set theory. Some important features of the nonmonotonic connectives are also discussed.

Development of Commonsense Knowledge Modeling System for Psychological Assessment in Clinical Psycho

According to the Buddhist philosophy, hatred （dosa） is considered as one of the three unwholesome roots which determine the actual immoral quality of volitional states and a conscious thought with its mental factors. Hatred, then, comprises all degrees of repulsion from the faintest trace of ill-humour up to the highest pitch of hate and wrath. Thus, ill-will, evil intention, wickedness, corruption and malice are various expressions and degrees ofdosa. A hateful temperament is said to be due to a predominance of the type of dosa, apo, vayu and semha. Vedic psychology forms the clinical core of mental health counseling in the Ayurvedic medical tradition. According to Ayurvedic medical practises, a person is dominated on one of constitutes type （type ofdosa） namely vata （vayu）, pita （apo） or kapha （semha）. This is known as prakurthi pariksha. Important aspect of identification of constitute type is for diagnosis of mental diseases, because each of constituent type has a list of probable mental diseases. An important area of expertise for many clinical psychologists is psychological assessment. Constructions of information systems using psychological assessment in clinical psychology have a problem of effective communication because of implicit knowledge. This complicates the effective communication of clinical data to the psychologist. In this paper, it presents an approach to modeling commonsense knowledge in clinical psychology in Ayurvedic medicine. It gives three-phase an approach for modeling commonsense knowledge in psychological assessment which enables holistic approach for clinical psychology. Evaluation of the system has shown 77% accuracy.

A Comprehensive Overview of CFN From a Commonsense Perspective

Chinese FrameNet(CFN)is a scenario commonsense knowledge base(CKB)that plays an important role in research on Chinese language understanding.It is based on the theory of frame semantics and English FrameNet(FN).The CFN knowledge base contains a wealth of scenario commonsense knowledge,including frames,frame elements,and frame relations,as well as annotated instances with rich scenario-related labels on Chinese sentences and discourses.In this paper,we conduct a comprehensive overview of CFN from a commonsense perspective,covering topics such as scenario commonsense representation,CFN resources,and its applications.We also summarize recent breakthroughs and identify future research directions.First,we introduce the concept of scenario commonsense,including its definitions,examples,and representation methods,with a focus on the relationship between scenario commonsense and the frame concept in CFN.In addition,we provide a comprehensive overview of CFN resources and their applications,highlighting the newly proposed frame-based discourse representation and a human-machine collaboration framework for expanding the CFN corpus.Furthermore,we explore emerging topics such as expanding the CFN resource,improving the interpretability of machine reading comprehension,and using scenario CKBs for text generation.

A Survey of Commonsense Knowledge Acquisition

Collecting massive commonsense knowledge （CSK） for commonsense reasoning has been a long time standing challenge within artificial intelligence research. Numerous methods and systems for acquiring CSK have been developed to overcome the knowledge acquisition bottleneck. Although some specific commonsense reasoning tasks have been presented to allow researchers to measure and compare the performance of their CSK systems, we compare them at a higher level from the following aspects： CSK acquisition task （what CSK is acquired from where）, technique used （how can CSK be acquired）, and CSK evaluation methods （how to evaluate the acquired CSK）. In this survey, we first present a categorization of CSK acquisition systems and the great challenges in the field. Then, we review and compare the CSK acquisition systems in detail. Finally, we conclude the current progress in this field and explore some promising future research issues.

Visual commonsense reasoning with directional visual connections

To boost research into cognition-level visual understanding,i.e.,making an accurate inference based on a thorough understanding of visual details,visual commonsense reasoning(VCR)has been proposed.Compared with traditional visual question answering which requires models to select correct answers,VCR requires models to select not only the correct answers,but also the correct rationales.Recent research into human cognition has indicated that brain function or cognition can be considered as a global and dynamic integration of local neuron connectivity,which is helpful in solving specific cognition tasks.Inspired by this idea,we propose a directional connective network to achieve VCR by dynamically reorganizing the visual neuron connectivity that is contextualized using the meaning of questions and answers and leveraging the directional information to enhance the reasoning ability.Specifically,we first develop a GraphVLAD module to capture visual neuron connectivity to fully model visual content correlations.Then,a contextualization process is proposed to fuse sentence representations with visual neuron representations.Finally,based on the output of contextualized connectivity,we propose directional connectivity to infer answers and rationales,which includes a ReasonVLAD module.Experimental results on the VCR dataset and visualization analysis demonstrate the effectiveness of our method.

A NOTE ON CONDITIONAL IMPLICATION

It is well known that there exists a tight connection between nonmonotonic reasoning and conditional implication. Many researchers have investigated it from various angles. Among th em, C.Boutilier and P.Lamarre have shown that some conditional implication may b e regarded as the homology of different nonmonotonic consequence relations. In t his paper, based on the plausibility space introduced by Friedman and Halpern, w e characterize the condition logic in which conditional implication is nonmonoto nic, and this result characterizes the conditional implication which may be rega rded as the corresponding object in Meta language for nonmonotonic inference rel ations.

Argumentation in disjunctive logic programming

Argumentation (abduction) is widely applied in artificial intelligence (AI) and law reasoning. However, the problem of how to perform argumentation in disjunctive logic programming (DLP) is still open.In addition, a unifying semantic framework is required for incorporating various semantics for DLP. An argumentation-theoretic framework for DLP by taking the disjuncts of negative literals as abducibles is presented. This semantics not only is a simple and intuitive framework for performing argumentation and abduction in DLP, but also provides a unifying framework for many key semantics of disjunctive logic programs. In particular, it is shown that the EGCWA, well-founded model and disjunctive stable models can all be embedded into this semantics.

A survey of script learning

Script is the structured knowledge representation of prototypical real-life event sequences.Learning the commonsense knowledge inside the script can be helpful for machines in understanding natural language and drawing commonsensible inferences.Script learning is an interesting and promising research direction,in which a trained script learning system can process narrative texts to capture script knowledge and draw inferences.However,there are currently no survey articles on script learning,so we are providing this comprehensive survey to deeply investigate the standard framework and the major research topics on script learning.This research field contains three main topics:event representations,script learning models,and evaluation approaches.For each topic,we systematically summarize and categorize the existing script learning systems,and carefully analyze and compare the advantages and disadvantages of the representative systems.We also discuss the current state of the research and possible future directions.