Evaluating automated and manual acquisition of anaphora resolution strategies

Abstract

We describe one approach to build an automatically trainable anaphora resolution system. In this approach, we use Japanese newspaper articles tagged with discourse information as training examples for a machine learning algorithm which employs the C4.5 decision tree algorithm by Quinlan (Quinlan, 1993). Then, we evaluate and compare the results of several variants of the machine learning-based approach with those of our existing anaphora resolution system which uses manually-designed knowledge sources. Finally, we compare our algorithms with existing theories of anaphora, in particular, Japanese zero pronouns.

Hybrid Approach to Pronominal Anaphora Resolution in English Newspaper Text

“…It provides a mechanism for coordinating the application of context-independent and context-dependent constraints and preferences for accurate partitioning of noun phrases into co-reference equivalence classes. [4] In 2001, Mitkov represented that the comparative evaluation for resolving anaphora has to be performed using the same pre-processing tools and on the same set of data. They proposed an evaluation environment for comparing anaphora resolution algorithms which is illustrated by presenting the results of the comparative evaluation on the basis of several evaluation measures.…”

“…Traditional approaches are mainly works in basic three steps as: (1) Deciding search limit or anaphoric accessibility space, (2) apply various constraints, and then (3) apply preferences. [4] A. Search limit or Anaphoric Accessibility Space:…”

Translation-Based Steganography

“…4 Wet paper codes [15] are a general mechanism that allows the sender to transmit a steganographic message without sharing the selection channel used to hide the information with the receiver. The fundamental idea behind wet paper codes is that the sender is only able to modify certain locations in the cover object -so-called dry spots.…”

Automatic Anaphora Resolution for Norwegian (ARN)

“…It started with a large-scale pattern-matching (Dagan and Itai 1990), but it is the unsupervised machine learning and probabilistic approaches that are today the area of most interest (Aone and Bennett 1995;Ge, Hale, and Charniak 1998).…”