Named Entity Recognition using Cross-lingual Resources: Arabic as an Example

Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 1558–1567, Sofia, Bulgaria, August 4-9 2013. c⃝2013 Association for Computational Linguistics Named Entity Recognition using Cross-lingual Resources: Arabic as an Example Kareem Darwish Qatar Computing Research Institute Doha, Qatar kdarwish@qf.org.qa Abstract Some languages lack large knowledge bases and good discriminative features for Name Entity Recognition (NER) that can general- ize to previously unseen named entities. One such language is Arabic, which: a) lacks a capitalization feature; and b) has relatively small knowledge bases, such as Wikipedia. In this work we address both problems by in- corporating cross-lingual features and knowl- edge bases from English using cross-lingual links. We show that such features have a dramatic positive effect on recall. We show the effectiveness of cross-lingual features and resources on a standard dataset as well as on two new test sets that cover both news and microblogs. On the standard dataset, we achieved a 4.1% relative improvement in F- measure over the best reported result in the literature. The features led to improvements of 17.1% and 20.5% on the new news and mi- croblogs test sets respectively. 1 Introduction Named Entity Recognition (NER) is essential for a variety of Natural Language Processing (NLP) ap- plications such as information extraction. There has been a fair amount of work on NER for a variety of languages including Arabic. To train an NER sys- tem, some of the following feature types are typi- cally used (Benajiba and Rosso, 2008; Nadeau and Sekine, 2009): - Orthographic features: These features include capitalization, punctuation, existence of digits, etc. One of the most effective orthographic features is capitalization in English, which helps NER to gener- alize to new text of different genres. However, capi- talization is not very useful in some languages such as German, and nonexistent in other languages such as Arabic. Further, even in English social media, capitalization may be inconsistent. - Contextual features: Certain words are indica- tive of the existence of named entities. For example, the word “said” is often preceded by a named en- tity of type “person” or “organization”. Sequence labeling algorithms (ex. Conditional Random Fields (CRF)) can often identify such indicative words. - Character-level features: These features typ- ically include the leading and trailing letters of words. In some languages, these letters could pre- fixes and suffixes. Such features can be indicative or counter-indicative of the existence of named entities. For example, a word ending with “ing” is typically not a named entity, while a word ending in “berg” is often a named entity. - Part-of-speech (POS) tags and morphological features: POS tags indicate (or counter-indicate) the possible presence of a named entity at word level or at word sequence level. Morphological features can mostly indicate the absence of named entities. For example, Arabic allows the attachment of pronouns to nouns and verbs. However, pronouns are rarely ever attached to named entities. - Gazetteers: This feature checks the presence of a word or a sequence of words in …

Content truncated to 500 words