ISWC2019 Papers
From ISWC2019, an international conference on Semantic Web technology, one of the artificial intelligence technologies for handling knowledge information. In the previous article, we discussed ISWC2018.
ISWC2019 was held in Auckland (New Zealand), and similar to the previous ISWC2018, there were many reports on knowledge (graph) data and their application to Q&A systems, etc.
One report that caught our attention was “Logical Semantics Approach for Data Modeling in XBRL Taxonomies” (the technical areas of XBRL are Financial Reporting, Natural Language, and Taxonomy Modeling). He mentioned that XBRL’s future contribution to semantic technology will be in the utilization of large-scale data (including AI) in these technical areas), “Knowledge-based geospatial data integration and visualization with Semantic Web technologies” (a report on the implementation of geovisualization using Semantic Web technologies for geospatial information processing. The report describes the recent increase in demand for Spatial Data Infrastructure (SDI) in Europe and the U.S., and its contribution as a methodology for handling ontologies in geospatial space.) The report also includes “How to make latent factors interpretable by feeding Factorization machines with knowledge graphs” (a demonstration of the Hybrid Factorization Machine (kaHFM). It mentions the validation of how to initialize latent factors for the Factorization Machine using knowledge graphs to train interpretable models), “Summarizing News Articles using Question-and-Answer Pairs via Learning” (Google research presentation on the use of semantic technology in Q&A systems. The system mines questions from data associated with news stories and learns questions directly from the story content. This is the first demonstration of a learning-based approach to generating structured summaries of news stories with question-answer pairs to capture important and interesting aspects of a news story. The validation data is from the SQuAD dataset 2). and “Using a Knowledge Graph of Scenes to Enable the Search of Autonomous DrivingData” (describes a demonstration at Bosch using semantic technology. The intended use of the data is to provide large-scale data for automated driving technology. Benefits include improved capability to represent, integrate, and query automated driving data. It is also anticipated that data scientists and engineers within various projects and departments will be able to reuse data from each other’s applications), “Using Event Graph to Improve Question Answering in E-commerce Customer Service” (describing AliMe, an intelligent assistant that provides a question-and-answer service, which can answer more than 90% of the millions of questions per day. This session proposed an Event Graph that provides a reasoning mechanism to obtain accurate answers to the question types “why”, “wherefore”, “what if”, and “how next”. Events are properties of choices of situations that can happen, and the baseline knowledge graph is generated using WIKIDATA, DBpedia, YAGO, etc.), “Querying Enterprise Knowledge Graph With Natural Language “(Describes research on interactive interfaces to large enterprise knowledge graphs. He calls this mechanism Yugen (a deep learning-based interactive AI that answers user questions). They are already working on cold start challenges and domain-specific data detection challenges for enterprises, etc. Yugen is voice-based, so the advantage is that the cost of learning a specific query language is reduced), “Product Classification Using Microdata Annotations” (which described the task of automatically classifying products into universal categories using markup data published on the Web (in this case, RDF and Microdata). The challenge would be to handle the information needed for classification (e.g., treatment of individual websites, consistency across websites, site-specific product labels, etc.). This will be an example of using RDF as input data for deep learning), “Difficulty-controllable Multi-hop Question Generation From Knowledge Graphs” (neural network-based He explained multi-hop questioning. A multi-hop question is a question that can only be answered by hopping a node in the graph structure two or more times. As a solution, he mentioned that he had implemented an encoder-decoder model conditional on the difficulty level and was able to generate complex questions via a large knowledge graph. However, it was difficult to achieve and seemed to require future research), “QaldGen: Towards Microbenchmarking of Question Answering Systems Over Knowledge Graphs” (testing domain-specific Q&A systems He mentioned that micro-benchmarking is necessary because of the time and effort required to generate questions when trying to do so. QaldGen was proposed as a framework for generating questions useful for micro-benchmarking), etc.
Details are given below. Translated with www.DeepL.com/Translator (free version)
research track Decentralized Indexing over a Network of RDF Peers Datalog Materialisation in Distributed RDF Stores with Dynamic Data Exchange How to Make Latent Factors Interpretable by Feeding Factorization Machines with Knowledge Graphs Observing LOD Using Equivalent Set Graphs: It Is Mostly Flat and Sparsely Linked Optimizing Horn-SHIQ Reasoning for OBDA Using a KG-Copy Network for Non-goal Oriented Dialogues Canonicalizing Knowledge Base Literals Bag Semantics of DL-Lite with Functionality Axioms Validating Shacl Constraints over a Sparql Endpoint Mapping Factoid Adjective Constraints to Existential Restrictions over Knowledge Bases Mining Significant Maximum Cardinalities in Knowledge Bases HapPenIng: Happen, Predict, Infer—Event Series Completion in a Knowledge Graph Uncovering the Semantics of Wikipedia Categories Qsearch: Answering Quantity Queries from Text A Worst-Case Optimal Join Algorithm for SPARQL Knowledge Graph Consolidation by Unifying Synonymous Relationships Skyline Queries over Knowledge Graphs Detecting Influences of Ontology Design Patterns in Biomedical Ontologies Popularity-Driven Ontology Ranking Using Qualitative Features Incorporating Literals into Knowledge Graph Embeddings Extracting Novel Facts from Tables for Knowledge Graph Completion Difficulty-Controllable Multi-hop Question Generation from Knowledge Graphs Type Checking Program Code Using SHACL Decentralized Reasoning on a Network of Aligned Ontologies with Link Keys Ontology Completion Using Graph Convolutional Networks Non-parametric Class Completeness Estimators for Collaborative Knowledge Graphs—The Case of Wikidata Pretrained Transformers for Simple Question Answering over Knowledge Graphs Learning to Rank Query Graphs for Complex Question Answering over Knowledge Graphs THOTH: Neural Translation and Enrichment of Knowledge Graphs Entity Enabled Relation Linking SHACL Constraints with Inference Rules Query-Based Entity Comparison in Knowledge Graphs Revisited Anytime Large-Scale Analytics of Linked Open Data Absorption-Based Query Answering for Expressive Description Logics TransEdge: Translating Relation-Contextualized Embeddings for Knowledge Graphs Unsupervised Discovery of Corroborative Paths for Fact Validation RDF Explorer: A Visual SPARQL Query Builder Capturing Semantic and Syntactic Information for Link Prediction in Knowledge Graphs A Framework for Evaluating Snippet Generation for Dataset Search Summarizing News Articles Using Question-and-Answer Pairs via Learning Product Classification Using Microdata Annotations Truthful Mechanisms for Multi Agent Self-interested Correspondence Selection The KEEN Universe VLog: A Rule Engine for Knowledge Graphs ArCo: The Italian Cultural Heritage Knowledge Graph Making Study Populations Visible Through Knowledge Graphs LC-QuAD 2.0: A Large Dataset for Complex Question Answering over Wikidata and DBpedia SEO: A Scientific Events Data Model DBpedia FlexiFusion the Best of Wikipedia > Wikidata > Your Data The Microsoft Academic Knowledge Graph: A Linked Data Source with 8 Billion Triples of The RealEstateCore Ontology FoodKG: A Semantics-Driven Knowledge Graph for Food Recommendation BTC-2019: The 2019 Billion Triple Challenge Dataset Extending the YAGO2 Knowledge Graph with Precise Geospatial Knowledge The SEPSES Knowledge Graph: An Integrated Resource for Cybersecurity SemanGit: A Linked Dataset from git Squerall: Virtual Ontology-Based Access to Heterogeneous and Large Data Sources List.MID: A MIDI-Based Benchmark for Evaluating RDF Lists A Scalable Framework for Quality Assessment of RDF Datasets QaldGen: Towards Microbenchmarking of Question Answering Systems over Knowledge Graphs Sparklify: A Scalable Software Component for Efficient Evaluation of SPARQL Queries over Distributed RDF Datasets ClaimsKG: A Knowledge Graph of Fact-Checked Claims CoCoOn: Cloud Computing Ontology for IaaS Price and Performance Comparison In use track Semantically-Enabled Optimization of Digital Marketing Campaigns An End-to-End Semantic Platform for Nutritional Diseases Management VLX-Stories: Building an Online Event Knowledge Base with Emerging Entity Detection Personalized Knowledge Graphs for the Pharmaceutical Domain Use of OWL and Semantic Web Technologies at Pinterest An Assessment of Adoption and Quality of Linked Data in European Open Government Data Easy Web API Development with SPARQL Transformer Benefit Graph Extraction from Healthcare Policies Knowledge Graph Embedding for Ecotoxicological Effect Prediction Improving Editorial Workflow and Metadata Quality at Springer Nature A Pay-as-you-go Methodology to Design and Build Enterprise Knowledge Graphs from Relational Databases
In the next article, we will discuss ISWC2020.
コメント