1998 ExploitingDiverseKnowSourcesViaMEinNER
- (Borthwick et al., 1998) ⇒ Andrew Borthwick, John Sterling, Eugene Agichtein, Ralph Grishman. (1998). “Exploiting Diverse Knowledge Sources via Maximum Entropy in Named Entity Recognition.” In: Proceedings of the Sixth Workshop on Very Large Corpora.
Subject Headings: Supervised NER, Sequence Tagging, Maximum Entropy Models.
Notes
- It is one of the Seminal Papers on Supervised Sequence Segmentation.
- It used a Sequence Tegger.
- It uses Eric Sven Ristad. (1998). Maximum entropy modeling toolkit, release 1.6 beta. http://www.mnemonic.com/software/memt,
- It can be contrasted to
- HMM-based named-entity recognizer (Bikel et al., 1999)
- MEMM-based recognizer (McCallum et al., 2000a).
Cited by
2006
- (Sarawagi, 2006) ⇒ Sunita Sarawagi. (2006). “Efficient Inference on Sequence Segmentation Models.” In: Proceedings of the 23rd International Conference on Machine Learning (ICML 2006). doi:10.1145/1143844.1143944
- QUOTE: This limitation is partly addressed by expanding the label set — for example, a popular choice in named entity recognition tasks (NER) is the Begin-Continue-End-Unique-other (BCEUO) encoding of entity labels (Borthwick et al., 1998)
2000
- (McCallum et al., 2000a) ⇒ Andrew McCallum, Dayne Freitag, and Fernando Pereira. (2000). “Maximum Entropy Markov Models for Information Extraction and Segmentation.” In: Proceedings of ICML-2000.
- However, we know of no previous general method that combines the rich state representation of Markov models with the flexible feature combination of exponential models. The MENE named-entity recognizer (Borthwick, Sterling, Agichtein, & Grishman, 1998) uses an exponential model to label each word with a label indicating the position of the word in a labeled-entity class (start, inside, end or singleton), but the conditioning information does not include the previous label, unlike our model. Therefore, it is closer to our ME-Stateless model. It is possible that its inferior performance compared to an HMM-based named-entity recognizer (Bikel et al., 1999) may have similar causes to the corresponding weakness of ME-Stateless relative to FeatureHMM in our experiments — the lack of representation of sequential dependencies.
Quotes
Abstract
This paper describes a novel statistical named-entity (i.e. "proper name") recognition system built around a maximum entity framework. By working within the framework of maximum entropy theory and utilizing a flexible object-based architecture, the system is able to make use of an extraordinarily diverse range of knowledge sources in making its tagging decisions. These knowledge sources include capitalization features, lexical features, features indicating the current section of text (i.e. headline or main body), and dictionaries of single or multi-word terms. The purely statistical system contains no hand-generated patterns and achieves a result comparable with the best statistical systems. However, when combined with other hand-coded systems, the system achieves scores that exceed the highest comparable scores thus-far published.,
| Author | volume | Date Value | title | type | journal | titleUrl | doi | note | year | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1998 ExploitingDiverseKnowSourcesViaMEinNER | Ralph Grishman Eugene Agichtein Andrew Borthwick John Sterling | Exploiting Diverse Knowledge Sources via Maximum Entropy in Named Entity Recognition | Proceedings of the Sixth Workshop on Very Large Corpora | http://acl.ldc.upenn.edu/W/W98/W98-1118.pdf | 1998 |