PSNUS: Web People Name Disambiguation by Simple Clustering with Rich Features

Proceedings of the 4th International Workshop on Semantic Evaluations (SemEval-2007), pages 268–271, Prague, June 2007. c⃝2007 Association for Computational Linguistics PSNUS: Web People Name Disambiguation by Simple Clustering with Rich Features Ergin Elmacioglu1 Yee Fan Tan2 Su Yan1 Min-Yen Kan2 Dongwon Lee1 1The Pennsylvania State University, USA 2National University of Singapore, Singapore {ergin,syan,dongwon}@psu.edu, {tanyeefa,kanmy}@comp.nus.edu.sg Abstract We describe about the system description of the PSNUS team for the SemEval-2007 Web People Search Task. The system is based on the clustering of the web pages by us- ing a variety of features extracted and gen- erated from the data provided. This system achieves Fα=0.5 = 0.75 and Fα=0.2 = 0.78 for the final test data set of the task. 1 Introduction We consider the problem of disambiguating person names in a Web searching scenario as described by the Web People Search Task in SemEval 2007 (Ar- tiles et al., 2007). Here, the system receives as in- put a set of web pages retrieved from a search en- gine using a given person name as a query. The goal is to determine how many different people are rep- resented for that name in the input web pages, and correctly assign each namesake to its corresponding subset of web pages. There are many challenges towards an effective solution. We are to correctly estimate the number of namesakes for a given person name and group doc- uments referring to the same individual. Moreover, the information sources to be processed are unstruc- tured web pages and there is no certain way of cor- rectly establishing a relation between any two web pages belonging to the same or different individuals. We have taken several approaches to analyze dif- ferent sources of information provided with the in- put data, and also compared strategies to combine these individual features together. The configuration that achieved the best performance (which were sub- mitted for our run) used a single named entity fea- ture as input to clustering. In the remainder of this paper, we first describe our system in terms of the clustering approach used and alternative features in- vestigated. We then analyze the results on the train- ing set before concluding the paper. 2 Clustering Algorithm Clustering is the key part for such a task. We have chosen to view the problem as an unsupervised hard clustering problem. First, we view the problem as unsupervised, using the training data for parameter validation, to optimally tune the parameters in the clustering algorithm. Secondly, we observed that the majority of the input pages reference a single indi- vidual, although there are a few that reference mul- tiple individuals sharing the same name. Hence, we view the problem as hard clustering, assigning input pages to exactly one individual, so that the produced clusters do not overlap. Hard clustering algorithms can be classified as ei- ther partitive or hierarchical. Agglomerative hierar- chical clustering generates a series of nested clusters by merging simple clusters into larger ones, while partitive methods try to find a pre-specified …

Content truncated to 500 words