Tag line generating system using knowledge extracted from statistical analyses

Hiroaki Yamane, Masafumi Hagiwara

Research output: Contribution to journalArticlepeer-review

2 Citations (Scopus)


This paper proposes a tag line generating system based on knowledge obtained by tag line analyses. Using a tag line corpus, two features of tag line—part-of-speech N-gram and word usage—are obtained as knowledge. Based on these features, the tag line generating system generates tag line candidates. More specifically, for a given theme, a keyword is entered into the proposed system. For each keyword, the system selects related terms unique to the theme. And enormous number of candidates is generated by using a large-scale N-gram corpus. The proposed system subsequently selects candidates based on the degree of similarity, grammatical structures, and mutual information volume. The highest scored candidates are selected as final output. The performance of the proposed system has been evaluated by subjects in terms of sentence quality, category appropriateness, keyword appropriateness, and overall quality. The experimental results show that the proposed system is able to generate suitable tag lines for a given theme and keywords, indicating a potential online tag line generator in the future.

Original languageEnglish
Pages (from-to)57-67
Number of pages11
JournalAI and Society
Issue number1
Publication statusPublished - 2013 Feb


  • Sentence generation
  • Statistical analysis
  • Tag line

ASJC Scopus subject areas

  • Philosophy
  • Human-Computer Interaction
  • Artificial Intelligence


Dive into the research topics of 'Tag line generating system using knowledge extracted from statistical analyses'. Together they form a unique fingerprint.

Cite this