Research Area: Text Analysis and Machine Learning
Description: Text Analysis and Machine Learning cover the research field known as Knowledge-based Systems. Here are some of the questions it tackles: knowing if we have sufficient knowledge to describe and solve a given task; representing knowledge in a manner adequate to the task in which it is to be used; acquiring and learning knowledge from examples; analyzing and understanding a variety of knowledge sources, particularly in the form of documents and texts expressed in languages such as English; and planning and organizing knowledge into actionable units (agents) that can tackle problems autonomously.
Applications: In addition to the applications mentioned above, this research area has applications in a variety of other areas (e.g., medicine, computer security, management, digital games, or fashion). For instance, we work on a system that will assist the Children’s Hospital of Eastern Ontario emergency room personnel in predicting the severity of asthma attacks in patients suffering from this condition. In another application, we investigate inter-relationships between different body measurements to better understand the typical consumer, from a virtual tailor’s perspective – it turns out that different sets of body measurements are useful to characterize each clothing size.
Professors:
| Inkpen | natural language processing, lexical semantics, information retrieval, speech recognition and synthesis, and intelligent agents for the Semantic Web |
| Japkowicz | machine learning, data mining, performance evaluation methods for machine learning, computer security, environmental safety |
| Matwin | machine learning, data mining and text mining, applications, knowledge-based systems |
| Szpakowicz | natural language processing, text summarization, lexical resources, sentiment analysis |
| Turcotte (group) | applications of machine learning in bioinformatics, algorithm design |
| Tran | intelligent agents, multi-agent systems, reinforcement learning, electronic commerce, trust and reputation modeling, recommendation systems, architecture for mobile e-business |
| Viktor | long-term preservation of software dependent data; data mining of anthropometric databases containing relational, 2D and 3D data; 3D protein structure retrieval and similarity search |
Research groups involving several professors:
Leadership:
- Best results at the international competition in Information Retrieval in Cross-Language Speech Retrieval (Inkpen);
- OCRI nomination for Best Start-up of the Year for Distil Interactive, an Ottawa company that grew out of research in machine learning for digital games (Matwin);
- Promise Software Engineering Repository used to facilitate research around the world providing reusable data for experiments in software engineering (Sayyad Shirabad);
- Innovative Application Award for the system for virtual tailoring at the Principles of Knowledge Discover in Databases Innovative Conference (Viktor).
Some recent projects:
- Information retrieval from automatic speech transcripts; Automatic identification of cognates and false friends between French and English; Semantic similarity of words and texts; Cross-language synonym nuances (Inkpen; funded by NSERC);
- Using unsupervised learning for network alert correlation (Japkowicz; funded by Defence Research & Development Canada); Machine learning for computer virus detection; Data mining methods for monitoring the Nuclear Test Ban Treaty (funded by Health Canada);
- Building profiles of digital game players using machine learning; Partial automation of systematic reviews of medical literature using automatic text classification (Matwin; funded by OCE, NSERC, Distil, Inc., and TrialStat!, Inc.);
- Summarizing short stories, Enhancing Roget's Thesaurus, Analyzing the Language of Negotiations, Recognizing emotions in text (Szpakowicz; funded by NSERC and SSHRC)
- Machine learning as a tool for Empirical Software Engineering: prediction of relationships of interest between components of large software systems; identification of useful software properties (Matwin and Sayyad Shirabad; funded by NSERC);
- Modeling trust and reputation in e-commerce; Designing e-commerce recommender systems (Tran; funded by NSERC and ORNEC)
- Simultaneous alignment and structure prediction of RNAs; secondary structure motifs for RNA sequences (Turcotte)
- Virtual garment tailoring through cluster analysis and classification (Viktor; funded by NSERC, CFI, OIT, ORNEC and IBM Canada)
Slide Show
Opportunities for collaboration: Opportunities for joint research exist at all levels, from exploratory research to contract research, from sponsoring a student to supporting a dedicated project, and many variations in between. University-based research can be very cost-effective for your company and in most cases your support can be used to leverage additional grants from government agencies resulting in a multiplying effect. To find out more about a specific lab/group, project, or to discuss your research needs, contact the Technology & Research Development Office at (613) 562 5800 x2440. Email: research@eng.uottawa.ca , or the researcher directly (The School of EECS).
Printable Handout 
Research Area Coordinator: Matwin