Topics & Ideas

A list of topic ideas for senior thesis, practicum or individual student research projects.

[ NLP | Social Robotics | CS Education ]

Natural Language Processing

Open source dialogue system

We have an open source dialogue system, using a finite state representation of dialogue and google speech recognition. Now we want to create a working dialogue system which can facilitate conversation over department web pages.


Union information retrieval system

Use language processing techniques to build a retrieval system over the Union course catalogue. Allow for dialogue, for refining queries, based on retrieved answers, or build a recommender system, using prior courses, or topics of interest. For basics of IR, see the first few chapters of:

http://nlp.stanford.edu/IR-book/html/htmledition/irbook.html


Build a better chatbot

Look at using augmented transition networks for ways to connect topics and issues in a chatbot. The bot should be able to converse about a number of topics, and gracefully shift between topics, returning to incomplete topics at the appropriate time. See:

Wilks, Y., N. Webb, A. Setzer, M. Hepple and R. Catizone
Machine Learning Approaches to Human Dialogue Modelling, in Kuppervelt, Dybkjaer and Bernsen (eds.) Advances in Natural Multimodal Dialogue Systems, Kluwer, Berlin, 2005.


Opinion mining

Use learning to determine opinions on topics in text, or twitter, using simple sentiment analysis as a start point, progressing to an understanding of linguistic units, sarcasm and the use of hashtags. See:

Lin, C., Z. Akcam, S. Shaikh, S. Small, K. Stahl, T. Strzalkowski and N. Webb
Revealing Contentious Concepts Across Social Groups, in Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC), pp 2838 – 2841. 2012.

D. Maynard, K. Bontcheva, D. Rout. Challenges in developing opinion mining tools for social media. In Proceedings of @NLP can u tag #usergeneratedcontent?! Workshop at LREC 2012, May 2012, Istanbul, Turkey.


Grammars of dialogue

Segment dialogue into topics, and use dialogue acts to determine structure of interaction around those topics. The goal being to see if, using topic structure, we can find reliable, repeated segments of dialogue that we might call things like ‘exchange’, or ‘information-request’ or ‘clarification’. See:

Field, D., S. Worgan, N. Webb, M. Hepple and Y. Wilks
Automatic Induction of Dialogue Structure from the Companions Dialogue Corpus, in Proceedings of the 4th International Workshop on Human-Computer Conversation, Bellagio, Italy, October 2008.

Lauri Carlson
Focus and Dialogue Games, in Cognitive Constraints on Communication, Synthese Language Library Volume 18, 1984, pp 295-333


Social Robotics

Mapping and navigation

Approaching people to ask for help

Attracting attention


CS Education

Python robotics exercises

Taking the existing social robotics workshop material, create a series of programming exercises using Python and the NXT robots, to be given to students from grades 10 to 12. See:

http://www.albany.edu/robotics/srw/