Laboratories - Natural Language Processing
Human Language Technologies Laboratory
100 Elkton Road, Professors Sandra Carberry and Kathy McCoy.
The Human Language Technology Laboratory is an umbrella for two language-related laboratories: disabilities technology and discourse. The laboratory closely collaborates with the Statistical Information Retrieval Laboratory and the Text Mining Laboratory.
The Disabilities Technology Laboratory, directed by Kathy McCoy, develops intelligent interfaces for people with disabilities that affect their ability to communicate. The ICICLE System is an intelligent English grammar checker and tutor for people who are deaf. Other projects assist people who have special communication needs. We make "talking with" a computer faster and more natural. Another project is to help a person who has visual impairments "scan" a text to find the area relevant to answering a question.
The Discourse Laboratory, under the direction of Sandra Carberry, addresses problems related to discourse and dialogue. The Graphs project treats information graphics (bar charts, line graphs, etc.) as a form of discourse with a communicative goal. We are applying language understanding and generation techniques to index, store, and retrieve graphics from a digital library, to develop an interactive dialogue system that conveys the content of graphics via speech to individuals with sight impairments, and to develop an interactive graph design assistant that will critique graphs with the objective of improving them so that they achieve their communicative goal. Current collaborators include Dr. Stephanie Elzer (Millersville University), Dr. Dan Chester, and graduate students.
Statistical Information Retrieval Laboratory
77/79 East Delaware Ave, Professor Ben Carterette.
The Statistical Information Retrieval Laboratory pursues novel models of information organization, storage, access, retrieval, and integration using statistical and information-theoretic approaches. One of the key problems in developing such models is that optimizing and evaluating their utility requires human input. We aim to minimize the human cost, or to accomplish much more with an allotted cost, thereby allowing research and development to proceed much faster.
Text Mining Laboratory
102 Smith Hall, Professor Vijay Shanker.
The Text Mining Laboratory, directed by Vijay Shanker, is concerned with the development of language technology algorithms to assist scientists to rapidly access relevant information from research literature. Projects include the extraction of targeted information, retrieval of relevant textual passages, and assistance in the knowledge discovery process. A related project involves rapid adaptation of language processing tools that were developed for a general domain to be used in a specific domain. A third project involves multi-disciplinary effort that integrates natural language cues found in large software programs and program analysis for multiple software development and maintenance tasks.