F4 Information retrieval meets human language technology
     Marko Tadic, Faculty of Philosophy, Zagreb

  Abstract   Full Paper html pdf   Presentation   Back to Program    

The paper tries to highlight several points where informational retrieval could benefit from human language technology. Although there is an increasing quantity of multimedia content on the web, the majority of information is still coded in natural languages(s). Since the global trend in web language usage is coding in native language, it leads to the situation in which web pages in English do not represent the majority of web-text anymore. To retrieve information from non-English pages requires several tools, which have to deal with particular human language. These tools can be borrowed from the field of human language technology (HLT) and applied to web-text. Some possible areas of HLT, which could be used in information retrieval from web-text, are:
- morphological processing: should be able to cope with different word-forms in particular language in order to make language specific full-text search available
- named entities recognition: should give the possibility to get information on concepts which have fixed names/ titles/formulas (such as personal/institutional/geographical names, temporal expressions etc.)
- semantic thesauruses: should give the ability to retrieve information on the basis of language synonymy/proximity
- machine (aided) translation: should give (rough) translations of web-text.