Applying Machine Learning for High‐Performance Named‐Entity Extraction

Abstract

This paper describes a machine learning approach to building an efficient and accurate name spotting system. Finding names in free text is an important task in many text-based applications. Most previous approaches were based on hand-crafted modules encoding language and genre-specific knowledge. These approaches had at least two shortcomings: they required large amounts of time and expertise to develop and were not easily portable to new languages and genres. This paper describes an extensible system that automatically combines weak evidence from different, easily available sources: parts-of-speech tags, dictionaries, and surface-level syntactic information such as capitalization and punctuation. Individually, each piece of evidence is insufficient for robust name detection. However, the combination of evidence, through standard machine learning techniques, yields a system that achieves performance equivalent to the best existing hand-crafted approaches.

Data Mining for Problem Discovery

“…For example, they can extract peoples names, locations, vehicle types, and accidents from a passage. Information extraction techniques can be rule-based Ciravegna et al (1999) and Krupka & Hausman (1998), statistics-based Witten et al (1999), or use machine learning Baluja et al (1999). Text categorization and clustering are like categorization and clustering performed in data mining, but performed on narratives or text.…”

Adaptive information extraction

“…Most of these improvements are based on the use of empirical methods in NLP. Given the kinds of knowledge needed by empirical approaches to NLP, machine learning techniques have been widely used for its acquisition: Sekine et al [1998]; Borthwick et al [1998]; Baluja et al [1999]; Borthwick [1999]; Takeuchi and Collier [2002]; Yarowsky [2003] for NE recognition, Cardie et al [2000] for chunking, and McCarthy and ; Aone and Bennet [1996]; Cardie and Wagstaff [1999]; Mitkov [1998]; Ng and Cardie [2003] for coreference resolution. Detailed thorough surveys on the use of ML techniques for NLP tasks can be also found Young and Bloothooft [1997]; Manning and Schütze [1999]; Mooney and Cardie [1999].…”

Token Identification Using HMM and PPM Models

“…While the two systems are similar, there are significant differences between them. Another system using decision trees is proposed by Baluja et al [3]. Like the systems described by both Sekine and Bennett et al, they utilized a part-of-speech tagger, dictionary lookups, and word-level features, such as all-uppercase, initial-caps, single-character, and punctuation features.…”