Piggyback: Using Search Engines for Robust Cross-Domain Named Entity Recognition

Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics, pages 965–975, Portland, Oregon, June 19-24, 2011. c⃝2011 Association for Computational Linguistics Piggyback: Using Search Engines for Robust Cross-Domain Named Entity Recognition Stefan R¨ud Institute for NLP University of Stuttgart Germany Massimiliano Ciaramita Google Research Z¨urich Switzerland Jens M¨uller and Hinrich Sch¨utze Institute for NLP University of Stuttgart Germany Abstract We use search engine results to address a par- ticularly difficult cross-domain language pro- cessing task, the adaptation of named entity recognition (NER) from news text to web queries. The key novelty of the method is that we submit a token with context to a search engine and use similar contexts in the search results as additional information for correctly classifying the token. We achieve strong gains in NER performance on news, in-domain and out-of-domain, and on web queries. 1 Introduction As statistical Natural Language Processing (NLP) matures, NLP components are increasingly used in real-world applications. In many cases, this means that some form of cross-domain adaptation is neces- sary because there are distributional differences be- tween the labeled training set that is available and the real-world data in the application. To address this problem, we propose a new type of features for NLP data, features extracted from search en- gine results. Our motivation is that search engine results can be viewed as a substitute for the world knowledge that is required in NLP tasks, but that can only be extracted from a standard training set or pre- compiled resources to a limited extent. For example, a named entity (NE) recognizer trained on news text may tag the NE London in an out-of-domain web query like London Klondike gold rush as a location. But if we train the recognizer on features derived from search results for the sentence to be tagged, correct classification as person is possible. This is because the search results for London Klondike gold rush contain snippets in which the first name Jack precedes London; this is a sure indicator of a last name and hence an NE of type person. We call our approach piggyback and search result- derived features piggyback features because we pig- gyback on a search engine like Google for solving a difficult NLP task. In this paper, we use piggyback features to ad- dress a particularly hard cross-domain problem, the application of an NER system trained on news to web queries. This problem is hard for two reasons. First, the most reliable cue for NEs in English, as in many languages, is capitalization. But queries are generally lowercase and even if uppercase char- acters are used, they are not consistent enough to be reliable features. Thus, applying NER systems trained on news to web queries requires a robust cross-domain approach. News to queries adaptation is also hard because queries provide limited context for NEs. In news text, the first mention of a word like Ford is often a fully qualified, unambiguous name like Ford Mo- tor Corporation or Gerald Ford. …

Content truncated to 500 words