Developing Multilingual Text Mining Workflows in UIMA and U-Compare
- Cite this paper as:
- Kontonasios G., Korkontzelos I., Ananiadou S. (2012) Developing Multilingual Text Mining Workflows in UIMA and U-Compare. In: Bouma G., Ittoo A., Métais E., Wortmann H. (eds) Natural Language Processing and Information Systems. NLDB 2012. Lecture Notes in Computer Science, vol 7337. Springer, Berlin, Heidelberg
We present a generic, language-independent method for the construction of multilingual text mining workflows. The proposed mechanism is implemented as an extension of U-Compare, a platform built on top of the Unstructured Information Management Architecture (UIMA) that allows the construction, comparison and evaluation of interoperable text mining workflows. UIMA was previously supporting strictly monolingual workflows. Building multilingual workflows exhibits challenging problems, such as representing multilingual document collections and executing language-dependent components in parallel. As an application of our method, we develop a multilingual workflow that extracts terms from a parallel collection using a new heuristic. For our experiments, we construct a parallel corpus consisting of approximately 188.000 PubMed article titles for French and English. Our application is evaluated against a popular monolingual term extraction method, C Value.
Keywordstext mining multilingual text mining workflows UIMA U-Compare multilingual term extraction
Unable to display preview. Download preview PDF.