UCD-PN: Selecting General Paraphrases Using Conditional Probability

Proceedings of the 5th International Workshop on Semantic Evaluation, ACL 2010, pages 234–237, Uppsala, Sweden, 15-16 July 2010. c⃝2010 Association for Computational Linguistics UCD-PN: Selecting General Paraphrases Using Conditional Probability Paul Nulty University College Dublin Dublin, Ireland paul.nulty@ucd.ie Fintan Costello University College Dublin Dublin, Ireland fintan.costello@ucd.ie Abstract We describe a system which ranks human- provided paraphrases of noun compounds, where the frequency with which a given paraphrase was provided by human volun- teers is the gold standard for ranking. Our system assigns a score to a paraphrase of a given compound according to the num- ber of times it has co-occurred with other paraphrases in the rest of the dataset. We use these co-occurrence statistics to com- pute conditional probabilities to estimate a sub-typing or Is-A relation between para- phrases. This method clusters together paraphrases which have similar meanings and also favours frequent, general para- phrases rather than infrequent paraphrases with more specific meanings. 1 Introduction SemEval 2010 Task 9, “Noun Compound Inter- pretation Using Paraphrasing Verbs”, requires sys- tems to rank paraphrases of noun compounds according to which paraphrases were most fre- quently produced for each compound by human annotators (Butnariu et al., 2010). This paper de- scribes a system which ranks a paraphrase for a given compound by computing the probability of the paraphrase occurring given that we have previ- ously observed that paraphrase co-occurring with other paraphrases in the candidate paraphrase list. These co-occurrence statistics can be built using either the compounds from the test set or the train- ing set, with no significant difference in results. The model is informed by two observations: people tend to use general, semantically light para- phrases more often than detailed, semantically heavy ones, and most paraphrases provided for a specific compound indicate the same interpreta- tion of that compound, varying mainly according to level of semantic detail. Given these two properties of the data, the ob- jective of our system was to test the theory that conditional probabilities can be used to estimate a sub-typing or Is-A relation between paraphrases. No information about the compounds was used, nor were the frequencies provided in the training set used. 2 Motivation Most research on the disambiguation of noun com- pounds involves automatically categorizing the compound into one of a pre-defined list of seman- tic relations. Paraphrasing compounds is an alter- native approach to the disambiguation task which has been explored by (Lauer, 1995) and (Nakov, 2008). Paraphrases of semantic relations may be verbs, prepositions, or “prepositional verbs” like found in and caused by. (Lauer, 1995) catego- rized compounds using only prepositions. (Nakov, 2008) and the current task use only verbs and prepositional verbs, however, many of the para- phrases in the task data are effectively just prepo- sitions with a copula, e.g. be in, be for, be of. The paraphrasing approach may be easier to integrate into applications such as translation, query-expansion and question-answering — its output is a set of natural language phrases rather than an abstract relation category. Also, most …

Content truncated to 500 words