Emmanuel Navarro

Wiktionary and NLP: improving synonymy networks

Wiktionary, a satellite of the Wikipedia initiative, can be seen as a potential resource for Natural Language Processing. It requires however to be processed before being used efficiently as an NLP resource. After describing the relevant aspects of Wiktionary for our purposes, we focus on its structural properties. Then, we describe how we extracted synonymy networks from this resource. We provide an in-depth study of these synonymy networks and compare them to those extracted from traditional resources. Finally, we describe two methods for semi-automatically improving this network by adding missing relations: (i) using a kind of semantic proximity measure; (ii) using translation relations of Wiktionary itself.

Semi-automatic endogenous enrichment of collaboratively constructed lexical resources: piggybacking onto wiktionary

The lack of large-scale, freely available and durable lexical resources, and the consequences for NLP, is widely acknowledged but the attempts to cope with usual bottlenecks preventing their development often result in dead-ends. This article introduces a language-independent, semi-automatic and endogenous method for enriching lexical resources, based on collaborative editing and random walks through existing lexical relationships, and shows how this approach enables us to overcome recurrent impediments. It compares the impact of using different data sources and similarity measures on the task of improving synonymy networks. Finally, it defines an architecture for applying the presented method to Wiktionary and explains how it has been implemented.

Semi-automatic enrichment of crowdsourced synonymy networks: the WISIGOTH system applied to Wiktionary

Semantic lexical resources are a mainstay of various Natural Language Processing applications. However, comprehensive and reliable resources are rare and not often freely available. Handcrafted resources are too costly for being a general solution while automatically-built resources need to be validated by experts or at least thoroughly evaluated. We propose in this paper a picture of the current situation with regard to lexical resources, their building and their evaluation. We give an in-depth description of Wiktionary, a freely available and collaboratively built multilingual dictionary. Wiktionary is presented here as a promising raw resource for NLP. We propose a semi-automatic approach based on random walks for enriching Wiktionary synonymy network that uses both endogenous and exogenous data. We take advantage of the wiki infrastructure to propose a validation “by crowds”. Finally, we present an implementation called WISIGOTH, which supports our approach.

Clustering bipartite graphs in terms of approximate formal concepts and sub-contexts

The paper first offers a parallel between two approaches to conceptual clustering, namely formal concept analysis (augmented with the introduction of new operators) and bipartite graph analysis. It is shown that a formal concept (as defined in formal concept analysis) corresponds to the idea of a maximal bi-clique, while sub-contexts, which correspond to independent “conceptual worlds” that can be characterized by means of the new operators introduced, are disconnected sub-graphs in a bipartite graph. The parallel between formal concept analysis and bipartite graph analysis is further exploited by considering “approximation” methods on both sides. It leads to suggest new ideas for providing simplified views of datasets, taking also inspiration from the search for approximate itemsets in data mining (with relaxed requirements), and the detection of communities in hierarchical small worlds.

Clustering sets of objects using concepts-objects bipartite graphs

In this paper we deal with data stated under the form of a binary relation between objects and properties. We propose an approach for clustering the objects and labeling them with characteristic subsets of properties. The approach is based on a parallel between formal concept analysis and graph clustering. The problem is made tricky due to the fact that generally there is no partitioning of the objects that can be associated with a partitioning of properties. Indeed a relevant partition of objects may exist, whereas it is not the case for properties. In order to obtain a conceptual clustering of the objects, we work with a bipartite graph relating objects with formal concepts. Experiments on artificial benchmarks and real examples show the effectiveness of the method, more particularly the fact that the results remain stable when an increasing number of properties are shared between objects of different clusters.

Comparing and fusing terrain network information

Terrain networks (or complex networks) is a type of relational information that is encountered in many fields. In order to properly answer questions pertaining to the comparison or to the merging of such networks, a method that takes into account the underlying structure of graphs is proposed. The effectiveness of the method is illustrated using real linguistic data networks and artificial networks, in particular.