RI: SMALL: Statistical Linguistic Typology

点击翻译按钮获取中文摘要

This project considers the unification of two view of language: thatfrom natural language processing and that from linguistic typology.Our view is that typological information is both useful for solvingreal-world natural language processing thats and automaticallyderivable from language data. This research first explores how to usetypological knowledge to improve performance on problems such asdependency parsing and machine translation for low density langauges.Intuitively, our statistical models waste time exploring a hypothesisspace that is too big: the space of realistic grammars is much smallerthan the space of all grammars. The second part of this researchconsiders the automatic acquisition and boostrapping of typologicalknowledge from raw text. The outcome of this research is: (a)improved statistical models for hard natural language processingproblems; and (b) a larger library of typological universals that havebeen derived automatically from data. Our outcomes are empiricallyevaluated on the raw language processing tasks and in terms of thequality of the universal implications mined from data, but comparingthem with known repositories of universals. Our results will impact the fields of natural language processing and linguistics. From the research side, this research will find applications in a wider variety of problems than the ones we intend to study; in particular, the use of linguistic universals in natural language processing technologywill fundamentally change the way multilinguality is addressed in thisfield. From a linguistics perspective, the goal of this project is toshed new light on linguistic universals. This should impact not onlythe area of typology, but also the study and preservation ofendangered languages. By automatically identifying typologicalfeatures and implications from data, the process of documentingendangered languages could be made more efficient: leading to asmaller loss of knowledge of these languages.