International audience

We propose the use of controlled natural language as a target for knowledge graph question answering (KGQA) semantic parsing via language models as opposed to using formal query languages directly. Controlled natural languages are close to (human) natural languages, but can be unambiguously translated into a formal language such as SPARQL. Our research hypothesis is that the pre-training of large language models (LLMs) on vast amounts of textual data leads to the ability to parse into controlled natural language for KGQA with limited training data requirements. We devise an LLM-specific approach for semantic parsing to study this hypothesis. To conduct our study, we created a dataset that allows the comparison of one formal and two different controlled natural languages. Our analysis shows that training data requirements are indeed substantially reduced when using controlled natural languages, which is relevant since collecting and maintaining high-quality KGQA semantic parsing training data is very expensive and time-consuming.

Many graph pattern mining algorithms have been designed to identify recurring structures in graphs. The main drawback of these approaches is that they often extract too many patterns for human analysis. Recently, pattern mining methods using the Minimum Description Length (MDL) principle have been proposed to select a characteristic subset of patterns from transactional, sequential and relational data. In this paper, we propose an MDL-based approach for selecting a characteristic subset of patterns on labeled graphs. A key notion in this paper is the introduction of ports to encode connections between pattern occurrences without any loss of information. Experiments show that the number of patterns is drastically reduced. The selected patterns have complex shapes and are representative of the data.

The open nature of Knowledge Graphs (KG) often implies that they are incomplete. Link prediction consists in inferring new links between the entities of a KG based on existing links. Most existing approaches rely on the learning of latent feature vectors for the encoding of entities and relations. In general however, latent features cannot be easily interpreted. Rule-based approaches offer interpretability but a distinct ruleset must be learned for each relation, and computation time is difficult to control. We propose a new approach that does not need a training phase, and that can provide interpretable explanations for each inference. It relies on the computation of Concepts of Nearest Neighbours (CNN) to identify similar entities based on common graph patterns. Dempster-Shafer theory is then used to draw inferences from CNNs. We evaluate our approach on FB15k-237, a challenging benchmark for link prediction, where it gets competitive performance compared to existing approaches.

Updating existing knowledge bases is crucial to take into account the information that are regularly discovered. However, this is quite tedious and in practice Semantic Web data are rarely updated by users. This paper presents UTILIS, an approach to help users create and update objects in RDF(S) bases. While creating a new object, o, UTILIS searches for similar objects, found by applying relaxation rules to the description of o, taken as a query. The resulting objects and their properties serve as suggestions to expand the description of o.