Learning context-free grammars from partially structured examples

Yasubumi Sakakibara, Hidenori Muramatsu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

30 Citations (Scopus)


In this paper, we consider the problem of inductively learning context-free grammars from partially structured examples. A structured example is represented by a string with some parentheses inserted to indicate the shape of the derivation tree of a grammar. We show that the partially structured examples contribute to improving the efficiency of the learning algorithm. We employ the GA-based learning algorithm for context-free grammars using tabular representations which Sakakibara and Kondo have proposed previously [7], and present an algorithm to eliminate unnecessary nonterminals and production rules using the partially structured examples at the initial stage of the GA-based learning algorithm. We also show that our learning algorithm from partially structured examples can identify a context-free grammar having the in- tended structure and is more flexible and applicable than the learning methods from completely structured examples [5].

Original languageEnglish
Title of host publicationGrammatical Inference: Algorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings
PublisherSpringer Verlag
Number of pages12
ISBN (Print)9783540452577
Publication statusPublished - 2000
Externally publishedYes
Event5th International Colloquium on Grammatical Inference, ICGI 2000 - Lisbon, Portugal
Duration: 2000 Sep 112000 Sep 13

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISSN (Print)03029743
ISSN (Electronic)16113349


Other5th International Colloquium on Grammatical Inference, ICGI 2000

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science


Dive into the research topics of 'Learning context-free grammars from partially structured examples'. Together they form a unique fingerprint.

Cite this