Search

This archive contains AI-KG with additional 300K triples used in thepaper "Completing Scientific Facts in Knowledge Graphs of Research Concepts", accepted in IEEE Access.&nbsp

The continuous growth of scientific literature brings innovations and, at the same time, raises new challenges. One of them is related to the fact that its analysis has become difficult due to the high volume of published papers for which manual effort for annotations and management is required. Novel technological infrastructures are needed to help researchers, research policy makers, and companies to time-efficiently browse, analyse, and forecast scientific research. Knowledge graphs i.e., large networks of entities and relationships, have proved to be effective solution in this space. Scientific knowledge graphs focus on the scholarly domain and typically contain metadata describing research publications such as authors, venues, organizations, research topics, and citations. However, the current generation of knowledge graphs lacks of an explicit representation of the knowledge presented in the research papers. As such, in this paper, we present a new architecture that takes advantage of Natural Language Processing and Machine Learning methods for extracting entities and relationships from research publications and integrates them in a large-scale knowledge graph. Within this research work, we i) tackle the challenge of knowledge extraction by employing several state-of-the-art Natural Language Processing and Text Mining tools, ii) describe an approach for integrating entities and relationships generated by these tools, iii) show the advantage of such an hybrid system over alternative approaches, and vi) as a chosen use case, we generated a scientific knowledge graph including 109,105 triples, extracted from 26,827 abstracts of papers within the Semantic Web domain. As our approach is general and can be applied to any domain, we expect that it can facilitate the management, analysis, dissemination, and processing of scientific knowledge., Accepted for publication in Future Generation Computer Systems journal - Special Issue on Machine Learning and Knowledge Graphs

Science communication has a number of bottlenecks that include the rising number of published research papers and its non-machine-accessible and document-based paradigm, which makes the exploration, reading, and reuse of research outcomes rather inefficient. Recently, Knowledge Graphs (KG), i.e., semantic interlinked networks of entities, have been proposed as a new core technology to describe and curate scholarly information with the goal to make it machine readable and understandable. However, the main drawback of the use of such a technology is that researchers are asked to manually annotate their research papers and add their contributions within the KGs. To address this problem, in this paper we propose SCICERO, a novel KG generation approach that takes in input text from research articles and generates a KG of research entities. SCICERO uses Natural Language Processing techniques to parse the content of scientific papers to discover entities and relationships, exploits state-of-the-art Deep Learning Transformer models to make sense and validate extracted information, and uses Semantic Web best practices to formally represent the extracted entities and relationships, making the written content of research papers machine-actionable. SCICERO has been tested on a dataset of 6.7M papers about Computer Science generating a KG of about 10M entities. It has been evaluated on a manually generated gold standard of 3,600 triples that cover three Computer Science subdomains (Information Retrieval, Natural Language Processing, and Machine Learning) obtaining remarkable results.

In recent years, we saw the emergence of several approaches for producing machine-readable, semantically rich, interlinked description of the content of research publications, typically encoded as knowledge graphs. A common limitation of these solutions is that they address a low number of articles, either because they rely on human experts to summarize information from the literature or because they focus on specific research areas. In this paper, we introduce the Computer Science Knowledge Graph (CS-KG), a large-scale knowledge graph composed by over 350M RDF triples describing 41M statements from 6.7M articles about 10M entities linked by 179 semantic relations. It was automatically generated and will be periodically updated by applying an information extraction pipeline on a large repository of research papers. CS-KG is much larger than all comparable solutions and offers a very comprehensive representation of tasks, methods, materials, and metrics in Computer Science. It can support a variety of intelligent services, such as advanced literature search, document classification, article recommendation, trend forecasting, hypothesis generation, and many others. CS-KG was evaluated against a benchmark of manually annotated statements, yielding excellent results.

Moving towards the next generation of personalized learning environments requires intelligent approaches powered by analytics for advanced learning contexts with enriched digital content. Micro-Learning through Massive Open Online Courses is riding the wave of popularity as a novel paradigm for delivering short educational videos in small pre-organized chunks over time, so that learners can get knowledge in a manageable way. However, with the ever-increasing number of videos, it has become challenging to arrange and search them according to specific categories. In this paper, we get around the problem by bridging Learning Analytics and Cognitive Computing to analyze the content of large video collections, going over traditional term-based methods. We propose an efficient and effective approach to automatically classify a collection of educational videos on pre-existing categories which uses (i) a Speech-to-Text tool to get video transcripts, (ii) Natural Language Processing and Cognitive Computing methods to extract semantic concepts and keywords from video transcripts for their representation, and (iii) Apache Spark as Big Data technology for scalability. Several classifiers are trained on the feature vectors extracted by Cognitive Computing tools. Then, we compared our approach with other combinations of state-of-the-art feature types and classifiers over a large-scale dataset we collected from Coursera. Considering the experimental results, we expect our approach can facilitate the development of Learning Analytics tools powered by Cognitive Computing to support content managers on micro-learning video management while improving how learners search videos.

A huge number of scholarly articles published every day in different domains makes it hard for the experts to organize and stay updated with the new research in a particular domain. This study gives an overview of a new approach, HierClasSArt, for knowledge aware hierarchical classification of the scholarly articles for mathematics into a predefined taxonomy. The method uses combination of neural networks and Knowledge Graphs for better document representation along with the meta-data information. This position paper further discusses the open problems about incorporation of new articles and evolving hierarchies in the pipeline. Mathematics domain has been used as a use-case.

The amount of scientific literature continuously grows, which poses an increasing challenge for researchers to manage, find and explore research results. Therefore, the classification of scientific work is widely applied to enable the retrieval, support the search of suitable reviewers during the reviewing process, and in general to organize the existing literature according to a given schema. The automation of this classification process not only simplifies the submission process for authors, but also ensures the coherent assignment of classes. However, especially fine-grained classes and new research fields do not provide sufficient training data to automatize the process. Additionally, given the large number of not mutual exclusive classes, it is often difficult and computationally expensive to train models able to deal with multi-class multi-label settings. To overcome these issues, this work presents a preliminary Deep Learning framework as a solution for multi-label text classification for scholarly papers about Computer Science. The proposed model addresses the issue of insufficient data by utilizing the semantics of classes, which is explicitly provided by latent representations of class labels. This study uses Knowledge Graphs as a source of these required external class definitions by identifying corresponding entities in DBpedia to improve the overall classification.

Cultural heritage institutions store and digitize large amounts of multimedia data inside archives to make archival records findable by archivists, scientists, and general public. Cataloging standards vary from archive to archive and, therefore, the sharing and use of this data are limited. To solve this issue, linked open data (LOD) is rising as an essential paradigm to open and provide access to the archival resources. Archives which are opened to the world knowledge benefit from external connections by enabling the application of automated approaches to process archival records, helping all stakeholders to gain valuable insights. In this paper, we present the Archive Dynamics Ontology (ArDO) - an ontology designed for describing the hierarchical nature of archival multimedia data, as well as its application on the example of archival resources about the Weimar Republic. Furthermore, ArDO semantically organizes multimedia archival resources in form of texts, images, audios, and videos by representing the dynamics related to their classification over time. ArDO tracks the changes of a specific hierarchical classification schema referred to as systematics adopted to organize archival resources under semantically defined keywords.

By offering courses and resources, learning platforms on the Web have been attracting lots of participants, and the interactions with these systems have generated a vast amount of learning-related data. Their collection, processing and analysis have promoted a significant growth of learning analytics and have opened up new opportunities for supporting and assessing educational experiences. To provide all the stakeholders involved in the educational process with a timely guidance, being able to understand student's behavior and enable models which provide data-driven decisions pertaining to the learning domain is a primary property of online platforms, aiming at maximizing learning outcomes. In this workshop, we focus on collecting new contributions in this emerging area and on providing a common ground for researchers and practitioners (Website: https://mirkomarras.github.io/l2d-wsdm2021).

Today, increasing numbers of people are interacting online and a lot of textual comments are being produced due to the explosion of online communication. However, a paramount inconvenience within online environments is that comments that are shared within digital platforms can hide hazards, such as fake news, insults, harassment, and, more in general, comments that may hurt someone’s feelings. In this scenario, the detection of this kind of toxicity has an important role to moderate online communication. Deep learning technologies have recently delivered impressive performance within Natural Language Processing applications encompassing Sentiment Analysis and emotion detection across numerous datasets. Such models do not need any pre-defined hand-picked features, but they learn sophisticated features from the input datasets by themselves. In such a domain, word embeddings have been widely used as a way of representing words in Sentiment Analysis tasks, proving to be very effective. Therefore, in this paper, we investigated the use of deep learning and word embeddings to detect six different types of toxicity within online comments. In doing so, the most suitable deep learning layers and state-of-the-art word embeddings for identifying toxicity are evaluated. The results suggest that Long-Short Term Memory layers in combination with mimicked word embeddings are a good choice for this task.