Query and tag translation for Chinese-Korean cross-language social media retrieval
|Query and tag translation for Chinese-Korean cross-language social media retrieval|
|Author(s)||Wang Y.-C., Chen J.-T., Tsai R.T.-H., Hsu W.-L.|
|Published in||Proceedings of the 2011 IEEE International Conference on Information Reuse and Integration, IRI 2011|
|Keyword(s)||Unknown (Extra: Collaborative tagging, Korean language, Media content, Media files, Named entities, Search results, Social media, Software translators, Translation method, Wikipedia, YouTube, Information use, Linguistics, Metadata, Potassium, Query processing, User interfaces, Websites, Translation (languages))|
|Article||BASE, CiteSeerX, Google Scholar|
|Web||Ask, Bing, Google (PDF), Yahoo!|
|Download and mirrors|
|Local copy||Not available|
|Remote mirror(s)||Not available|
|Export and share|
|BibTeX, CSV, RDF, JSON|
|Browse properties · List of conference papers|
Query and tag translation for Chinese-Korean cross-language social media retrieval is a 2011 conference paper written in English by Wang Y.-C., Chen J.-T., Tsai R.T.-H., Hsu W.-L. and published in Proceedings of the 2011 IEEE International Conference on Information Reuse and Integration, IRI 2011.
Collaborative tagging has been widely adopted by social media websites to allow users to describe content with metadata tags. Tagging can greatly improve search results. We propose a cross-language social media retrieval system (CLSMR) to help users retrieve foreign-language tagged media content. We construct a Chinese to Korean CLSMR system that translates Chinese queries into Korean, retrieves content, and then translates the Korean tags in the search results back into Chinese. Our system translates NEs using a dictionary of bilingual NE pairs from Wikipedia and a pattern-based software translator which learns regular NE patterns from the web. The top-10 precision of YouTube retrieved results for our system was 0.39875. The K-C NE tag translation accuracy for the top-10 YouTube results was 77.6%, which shows that our translation method is fairly effective for named entities. A questionnaire given to users showed that automatically translated tags were considered as informative as a human-written summary. With our proposed CLSMR system, Chinese users can retrieve online Korean media files and get a basic understanding of their content with no knowledge of the Korean language.
- This section requires expansion. Please, help!
Probably, this publication is cited by others, but there are no articles available for them in WikiPapers.