<p>This thesis focuses on learning natural language interfaces using synchronous</p><p>grammars, l-calculus and statistical modeling of parse probabilities. A major</p><p>focus of the thesis has been to replicate Mooney and Wong’s l-WASP [17] algorithm</p><p>and implement it inside the C-PHRASE [12] Natural Language Interface</p><p>(NLI) system. By doing this we can use C-PHRASE’s more expressive and transportable</p><p><em>meaning representation language </em>(MRL), rather than the PROLOG-based</p><p>MRL Mooney and Wong used.</p><p>Our system, the C-PHRASE LEARNER, relaxes some constraints in l-WASP</p><p>to allow use of more flexible MRL grammars. We also reformulate the algorithm</p><p>in terms of operations on trees to clarify and simplify the approach. We test the</p><p>C-PHRASE LEARNER over the US geography corpus GEOQUERY and produce</p><p>precision and recall results slightly below those achieved by l-WASP. This was</p><p>expected as we have fewer domain restrictions due to our more expressive and</p><p>portable MRL grammar.</p><p>Our work on the C-PHRASE LEARNER system has also revealed some promising</p><p>avenues of future research including, among others, alternative statistical alignment</p><p>strategies, integrating linguistic theories into our learning algorithm and</p><p>ways to improve named entity recognition. C-PHRASE LEARNER is presented</p><p>as open source to the community to allow anyone to expand upon this work.</p>
| Identifer | oai:union.ndltd.org:UPSALLA/oai:DiVA.org:umu-35392 |
| Date | January 2010 |
| Creators | Granberg, Johan |
| Publisher | Umeå University, Department of Computing Science |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, text |
| Relation | UMNAD ; 842 |
Page generated in 0.0012 seconds