The project focuses on evaluation, which is the key to making progress in building better search and classification solutions. It involves the study of the adequacy of different types of Internet repositories as data sources to create test collections for research on early risk prediction and the design and construction of a new such collection. The collection will be made available to the research community, together with a suitable evaluation methodology.
Funding Agency: Swiss National Science Foundation (SNSF)
Starting date: June 2015
Duration: 6 months
This project aims at advancing research on opinion mining by looking at sentences expressing comparative opinions. The project objectives are: 1) develop a model for identifying comparative opinions; 2) implement a demonstration system of the above model in a realistic application scenario; 3) develop a proper evaluation framework for detecting and tracking emotion expressions implying comparative opinions over time; 4) finally, evaluate the proposed models with the evaluation framework developed in point 3. This will involve a close collaboration between the two teams of USI (Prof. Fabio Crestani) and the University of Malaya (Dr. Kasturi Dewi Varathan).
Funding Agency: Scientific & Technological Cooperation Programme
Starting date: June 2015
Duration: 6 months
The objective of query-rewriting is to rewrite the original query submitted by the user to better fit it to the perceived user information need. In this project we will design a set of algorithms for the personalisation and geo-contextualisation of query suggestion and rewriting. The expected results will be measured in terms of the correctness of the rewriting model, where we expect a substantial increase in performance compared with other personalisation (without contextualisation) models.
Starting date: Feb 2015
In this project we investigate searching and browsing in social sites and determine where the one-shot and one-size-fits-all paradigm of search is failing users and does not sufficiently assist them with their information gathering task. We use modern statistical learning techniques to develop models that are able to utilise personalisation, temporal task-based knowledge and topical information derived from the corpus to improve search. The proposed work significantly extend earlier work in personalisation of social media search and latent topic models carried out by the applicant.
This COST Action explores innovative frameworks to empower the synergies from the disparate research fields of Machine Translation, Information Retrieval and Multifaceted Interactive Information Access within the specific context of patent search and other next generation Web applications. See here for more information.
HEBE -Highly Engaging e-Book Experiences- aims at producing novel interfaces for playing, interacting and reading e-books for children. We propose in this study to involve children in exploring different type of technology, hardware and software, in order to produce more engaging, usable and fun e-book interfaces for them. The main hypothesis of the study is that in order to make e-reading a fun experience for children new innovative interfaces are needed and children should take an active role in their design.
This collaborative project with AT&T Labs deals with designing, implementing and evaluation new mobile interfaces for information retrieval. In particular, in this project we are developing a Just-in-Time Mobile Information Retrieval (JIT-MobIR) system. The system is able to predict user actions and needs and generate appropriate queries to find the information before the user actually knows he needs it.
The project has two different but integrated objectives. The first is to study how selected categories of users search for patents in relation to specific information needs and tasks. This will result in a set of guidelines to drive the design of innovative and more usable interactive systems for patent search. The second objective is to use these guidelines for the design and implementation of a new interactive patent search system that will then be evaluated with real users in the context of real search tasks, using a user and task oriented evaluation methodology. The emphasis in both objectives is on interactivity.
The vast majority of current TDT research makes two important assumptions about the nature of news: a) every news story discusses only one topic (in order to place a story to exactly one cluster); b) any hierarchical relationships between events should be ignored (to avoid the problem of events within events). The aim of this project is to devise some clustering technique that will allow us to remove (or significantly simplify) these two assumptions, as they limit considerably the power of TDT models.
In this project we will extend the latest models of statistical content analysis, that are proving successful in the areas of text mining and information retrieval, for the mining of conversational content for topic identification (what is the conversation about?) and author identification (who are the people involved in the conversation?).
The main objective of the project is to investigate multi-stream Topic Detection and Tracking by designing new algorithms for robust Story Link Detection in noisy streams that will work across several heterogeneous text streams. In addition, the project will study multi/cross-stream dependency, centrality, and authoritativeness, applying to streams work currently being carried out in stream segmentation and stream data mining.
In this project we investigate the combination of Soft Computing and statistical techniques for feature selection for the tasks of opinion ﬁnding and opinion polarity detection, in order to create automatically a weighted dictionary of opinion-bearing terms that will be used to determine, in a query independent way, the topical-opinion of a blog post. The project is closely connected to the IR group participation in the COST Action IC0702.
The main objective of the COST action IC0702 "Combining Soft Computing Techniques and Statistical Methods to Improve Data Analysis Solutions" is to strengthen the dialogue between the statistics and soft computing research communities in order to cross-pollinate both fields and generate mutual improvement activities. Here for more information.
The project extends and complements the work carried out in the SNSF project ADIR (Adaptive Information Retrieval), by designing, implementing and testing personalised and adaptive resource selection methods to be used for Distributed Information Retrieval with un-cooperative resources.
The project aims at designing, implementing and evaluating new tools for allowing easier interactions with users when measuring usability of interactive systems. In particular, the project will concentrate on tangible tool for children.
The project aims at the design, implementation, and evaluation of an interactive modus operandi mining and visualisation system, integrating behavioural and geographical information. The system it to be used by the police for suspect prioritisation and investigation Management. The project was based at the University of Strathclyde. More information here.
This FP7 STREP project will develop new interaction paradigms to allow children to easily express their information need, to have results presented in an intuitive way and to engage children in system interaction. It will also propose novel child-centered evaluation methodologies. The project is based at the University of Strathclyde.
The project is concerned withe the design, implementation and evaluation of personalised and adaptive resource description methods for un-cooperative resources. The produced resource descriptions will be used by advance resource selections methods for Distributed Information Retrieval.
The project is concerned with the design, implementation and testing of advanced resource description, resource selection and results fusions methods for Distributed Information Retrieval. These methods can be automatically personalised to the user task and user needs and are specifically designed to access information held in the Deep Web (i.e. in non-cooperative and heterogeneous resources). The project is based at the University of Strathclyde. Here for more information.
This project aims at investigating different mathematical models of cross-media and cross-modal indexing. In particular, it will develop models of heterogeneous evidence combination that will also be useful for other applications, like for example blog search and topic distillation.
The PENG project aims at defining a flexible, personalised and context-aware system for the gathering, filtering, retrieval and presentation of multimedia news for news professionals (e.g. journalists and editors), with a view of making the system also available later for general users. The project was based at the University of Strathclyde.Project website.
REVEAL-THIS addresses a basic need underlying content organisation, filtering, consumption and enjoyment by developing content programming systems that will help European citizens keep up with the explosion of digital content scattered over different platforms, media and different languages. The project was based at the University of Strathclyde. Project website
The objective of DILIGENT is to create an advanced test-bed that will allow members of dynamic virtual e-Science organisations to access shared knowledge and to collaborate in a secure, coordinated, dynamic and cost-effective way. This test-bed will be built by integrating Grid and Digital Library technologies. The project was based at the University of Strathclyde. Project website.
The project investigated a formalisation of the New Event Detection problem that fits into the PENG (Personalised News Content Programming) framework. This formalisation allows for a personal definition of what a new event is, obtained from user feedback. This enables for the definition of models of NED that are adaptive to the user changing interests and changing perception of what constitutes a new event. The project was based at the University of Strathclyde and funded a 6 months collaboration with Xerox Parc, in Paolo Alto, CA, USA. More info here.
The project investigated new algorithms for topic detection and tracking of news stories that enable to detect and take into account geographical and temporal references present in the news stories and in the user requests. The project was based at the University of Strathclyde and funded a 2-months visit to the University of California at Berkeley, CA, USA.
This project aims at designing new models for automatically detecting the size of a textual database or a digital library by query-sampling its content via its public query interface. The database size is an important parameter for resource selection algorithms used in Distributed Information Retrieval. The project was based at the University of Strathclyde.
The objective of MIND is to design models and to build sets of tools and associated test-beds to improve the effectiveness of resource selection, multimedia information access, retrieval and fusion of the retrieved data from non-cooperative distributed and heterogeneous multimedia digital libraries. The project was based at the University of Strathclyde. Project website.
WEBKIT utilises a Tangible User Interface for navigating sources of information (e.g. the WWW) that places priority on physical manipulation by the users. Essentially users control the system and navigate, manipulate and utilise data and information by selecting and positioning physical objects. The project was based at the University of Strathclyde. More information here.
DIOGENE aims at designing, implementing and evaluating an innovative training Web brokering environment for ICT individual training. This environment will be able to support learners during the whole cycle of the training, from the definition of objectives to the assessment of results. The project was based at the University of Strathclyde.More information here.