Using Wikipedia categories for compact representations of chemical documents

Abstract

Today, Web pages are usually accessed using text search engines, whereas documents stored in the deep Web are accessed through domain-specific Web portals. These portals rely on external knowledge bases, respectively ontologies, mapping documents to more general concepts allowing for suitable classifications and navigational browsing. Since automatically generated ontologies are still not satisfactory for advanced information retrieval tasks, most portals heavily rely on hand-crafted domain-specific ontologies. This, however, also leads to high creation and maintaining costs. On the other hand, a freely available community maintained, if somewhat general, knowledge base is offered by Wikipedia. During the last years the coverage of Wikipedia has reached a large pool of information including articles from almost all domains. In this paper, we investigate the use of Wikipedia categories to describe the content of chemical documents in a compact form. We compare the results to the domain-specific ChEBI ontology and the results show that Wikipedia categories indeed allow useful descriptions for chemical documents that are even better than descriptions from the ChEBI ontology.

Exploiting the wisdom of the crowds for characterizing and connecting heterogeneous resources

“…In fact, numerous previous works leverage the use of Wikipedia categories. For example, Köhncke and Balke [11] exploit Wikipedia categories in order to generate useful descriptions for chemical documents. In their work, they identify chemical entities in documents and extract the categories of these entities.…”

“…Additionally, categories are organized in a graph in which sub-categories reference to top-level categories. The English Wikipedia has a total of 23 top-level categories (Main topic classifications), which we use to represent a profile 11 . The creation of semantically enhanced profiles consists of three stages.…”

EUMSSI: a Platform for Multimodal Analysis and Recommendation using UIMA

“…Wikipedia categories have been successfully exploited with this purpose in different works: e.g. to describe chemical documents (Köhncke and Balke, 2010), to identify topics of interest for Twitter users (Michelson and Macskassy, 2010), and also to improve Web video categorization (Chen et al, 2010). Moreover, (Hahn et al, 2010) have shown that the structured information gathered from Wikipedia infoboxes can be used to answer complex questions, like "Which Rivers flow into the Rhine and are longer than 50 kilometers?"…”

The Heterogeneous Cluster Ensemble Method Using Hubness for Clustering Text Documents

“…They fail to capture other useful semantic knowledge, e.g. Wikipedia category that contains much meaningful information in the form of a hierarchical ontology [11]. In addition, these methods failed to model and cluster documents represented with multiple feature space (or relations).…”