Food computing is currently fast-growing into an innovative area of knowledge extraction. However, benchmarks for information extraction from semi-structured data, especially when dealing with more complex relations, are scarce in this domain. In this paper, we introduce a benchmark aimed at information extraction of complex entities to support ingredient substitution tasks. Firstly, we present a new dataset – called TASTEset – for fine-grained recognition of food entities in culinary recipes. Secondly, we provide complex entity annotations for substitution on top of the fine-grained entity mentions, which we carefully prepared. We share the dataset and the tasks to encourage progress on more in-depth and complex information extraction from recipes.

Abstract Recent years have shown that deep learning models pre-trained on large text corpora using the language model objective can help solve various tasks requiring natural language understanding. However, many commonsense concepts are underrepresented in online resources because they are too obvious for most humans. To solve this problem, we propose the use of affordances – common-sense knowledge that can be injected into models to increase their ability to understand our world. We show that injecting ConceptNet knowledge into BERT-based models leads to an increase in evaluation scores measured on the PIQA dataset.

Ontology authoring is a complicated and error-prone process since the knowledge being modeled is expressed using logic-based formalisms, in which logical consequences of the knowledge have to be foreseen. To make that process easier, competency questions (CQs), being questions expressed in natural language are often stated to trace both the correctness and completeness of the ontology at a given time. However, CQs have to be translated into a formal language, like ontology query language (SPARQL-OWL), to query the ontology. Since the translation step is time-consuming and requires familiarity with the query language used, in this paper, we propose an automatic method named SeeQuery, which recommends SPARQL-OWL queries being translations of CQs stated against a given ontology. It consists of a pipeline of transformations based on template matching and filling, being motivated by the biggest to date publicly available CQ to SPARQL-OWL datasets. We provide a detailed description of SeeQuery and evaluate the method on a separate set of 2 ontologies with their CQs. It is, to date, the only automatic method available for recommending SPARQL-OWL queries out of CQs. The source code of SeeQuery is available at: https://github.com/dwisniewski/SeeQuery.

The authors propose a new method for mining sets of patterns for classification, where patterns are represented as SPARQL queries over RDFS. The method contributes to so-called semantic data mining, a data mining approach where domain ontologies are used as background knowledge, and where the new challenge is to mine knowledge encoded in domain ontologies, rather than only purely empirical data. The authors have developed a tool that implements this approach. Using this the authors have conducted an experimental evaluation including comparison of our method to state-of-the-art approaches to classification of semantic data and an experimental study within emerging subfield of meta-learning called semantic meta-mining. The most important research contributions of the paper to the state-of-art are as follows. For pattern mining research or relational learning in general, the paper contributes a new algorithm for discovery of new type of patterns. For Semantic Web research, it theoretically and empirically illustrates how semantic, structured data can be used in traditional machine learning methods through a pattern-based approach for constructing semantic features.