Unsupervised graph-based topic labelling using dbpedia

Abstract

Automated topic labelling brings benefits for users aiming at analysing and understanding document collections, as well as for search engines targetting at the linkage between groups of words and their inherent topics. Current approaches to achieve this suffer in quality, but we argue their performances might be improved by setting the focus on the structure in the data. Building upon research for concept disambiguation and linking to DBpedia, we are taking a novel approach to topic labelling by making use of structured data exposed by DBpedia. We start from the hypothesis that words co-occuring in text likely refer to concepts that belong closely together in the DBpedia graph. Using graph centrality measures, we show that we are able to identify the concepts that best represent the topics. We comparatively evaluate our graph-based approach and the standard text-based approach, on topics extracted from three corpora, based on results gathered in a crowd-sourcing experiment. Our research shows that graph-based analysis of DBpedia can achieve better results for topic labelling in terms of both precision and topic coverage.

Path-Based Semantic Relatedness on Linked Data and Its Use to Word and Entity Disambiguation

“…Table 1 shows the exact number of pairs of words that we could directly and unambiguously link to concepts from DBpedia and Freebase. We remove from the graph of DBpedia the so-called stopURIs [12,27]. Regarding Freebase, we remove all edges with an exclusivity score lower than 10 −7 as they bring no impact on our measures due to their very small contribution, but they dramatically impact the performance of graph traversal algorithms.…”

Proceedings of the First AHA!-Workshop on Information Discovery in Text

“…A potential context of the content can likely be inferred by extracting set of topics that bound the text. Hulpus et al (2013) proposed an approach by linking the topics inherent to a text with the concepts in DBpedia 2 and thereby automatically extracting the topic labels from the corpus. Meij et al (2012) extracted underlying concepts of a text from a large knowledge base of Wikipedia articles by applying a supervised learning using a Naive Bayes (NB), Support Vector Machines (SVM), and a C4.5 decision tree classifier.…”

Kanopy: Analysing the Semantic Network around Document Topics

“…A semantic network is extracted by activating the concepts and their relations within a distance of two hops from each seed. In the majority of cases, this is sufficient to connect the single concept subgraphs into one larger graph, which provides the input for the next stage [3]. 5.…”

“…From this extracted topic graph we finally identify the most relevant concepts. We apply the focused random walk betweenness and focused information centrality, as defined in [3], which rank the nodes in the topic graph with respect to their semantic centrality to the seed concepts.…”