Dutch Word Sense Disambiguation: Data and Preliminary Results

Dutch Word Sense Disambiguation: Data and Preliminary Results Iris Hendrickx* and Antal van den Bosch*,+ * ILK / Computational Linguistics, Tilburg University, NL-5000 LE Tilburg, The Netherlands + WhizBang! Labs-Research, 4616 Henry Street, Pittsburgh PA 15213, USA Abstract We describe the Dutch word sense disambigua- tion data submitted to SENSEVAL-2, and give preliminary results on the data using a WSD system based on memory-based learning and statistical keyword selection. 1 Introduction Solving lexical ambiguity, or word sense disam- biguation (WSD), is an important task in Nat- ural Language Processing systems. Much like syntactic word-class disambiguation, it is not a end in itself, but rather a subtask of other nat- ural language processing tasks (Kilgarriff and Rozenzweig, 2000). The problem is far from solved, and research and competition in the de- velopment of WSD systems in isolation is mer- ited, preferrably on many different languages and genres. Here we introduce the first electronic Dutch word-sense annotated corpus, that was collected under a sociolinguistic research project (Sch- rooten and Vermeer, 1994), and was kindly do- nated by the team coordinators to the WSD systems community. In this paper we describe the original data and the preprocessing steps that were applied to it before submission to the SENSEVAL-2, in Section 2. We also present the first, preliminary, results obtained with MBWSD- D, the Memory-Based Word-Sense Disambigua- tion system for Dutch, that uses statistical key- word selection, in Section 3. 2 Data: The Dutch child book corpus The Dutch WSD corpus was built as a part of a sociolinguistic project, led by Walter Schrooten and Anne Vermeer (1994), on the active vocab- ulary of children in the age of 4 to 12 in the Netherlands. The aim of developing the corpus 13 was to have a realistic wordlist of the most com- mon words used at elementary schools. This wordlist was further used in the study to make literacy tests, including tests how many senses of ambiguous words were known by children of different ages. The corpus consists of texts of 102 illustrated children books in the age range of 4 to 12. Each word in these texts is manually annotated with its appropriate sense. The data was annotated by six persons who all processed a different part of the data. Each word in the dataset has a non- hierarchical, symbolic sense tag, realised as a mnemonic description of the specific meaning the word has in the sentence, often using a re- lated term. As there was no gold standard sense set of Dutch available, Schrooten and Vermeer have made their own set of senses. Sense tags consist of the word's lemma and a sense description of one or two words ( dro- gen_nat) or a reference of the grammatical cat- egory (fiets_N, fietsen_ V). Verbs have as their tag their lemma and often a reference to their function in the sentence (is/ zijn_kww). When a word has only one sense, this is represented with a simple"=". Names and sound imitations also have "=" …

Content truncated to 500 words