Henry Tirri

Multi-faceted information retrieval system for large scale email archives

We profile a system for search and analysis of large-scale email archives. The system builds around four facets: Content-based search engine, statistical topic model, automatically inferred social networks and time-series analysis. The facets correspond to the types of information available in email data. The presented system allows chaining or combining the facets flexibly. Results of one facet may be used as input to another, yielding remarkable combinatorial power. In information retrieval point of view, the system provides support for exploration, approximate textual searches and data visualization. We present some experimental results based on a large real-world email corpus.

Utilizing Rich Bluetooth Environments for Identity Prediction and Exploring Social Networks as Techniques for Ubiquitous Computing

Personal identification and using that information is in the heart of many ubiquitous systems. We present two complementary techniques, namely personal identification without directly observing the subject, and using that information for understanding the social relations between the subjects. We show that with certain presumptions it is possible to predict ones identity with reasonable certainty only by observing ones Bluetooth neighborhood without the need to directly observe the subject. We also show how this information can be used for exploring the social relations between the subjects

A temporally adaptive content-based relevance ranking algorithm

In information retrieval relevance ranking of the results is one of the most important single tasks there are. There are many diffierent ranking algorithms based on the content of the documents or on some external properties e.g. link structure of html documents.We present a temporally adaptive content-based relevance ranking algorithm that explicitly takes into account the temporal behavior of the underlying statistical properties of the documents in the form of a statistical topic model. more we state that our algorithm can be used on top of any ranking algorithm.

Multi-Faceted Information Retrieval System for Large Scale Email Archives

We profile a system for search and analysis of large-scale email archives. The system builds around four facets: content-based search engine, statistical topic model, automatically inferred social networks, and time-series analysis. The facets correspond to the types of information available in email data. The presented system allows chaining or combining the facets flexibly. Results of one facet may be used as input to another yielding remarkable combinatorial power. In information retrieval point of view, the system provides support for exploration, approximate textual searches and data visualization. We present some experimental results based on a large real-world email corpus.

Supporting open-ended discourse with transparent groupware

Web-based learning communities need tools that provide specific support to their progressive and open-ended efforts. The paper starts with an overview of some fundamental principles and phenomena underlying the activities of learning communities, and suggests openness and transparency as two crucial requirements for effective tools. In particular, learning communities should be enabled to utilise any available information sources effectively and engage in artifact-centred discourse. We introduce a system called OurWeb, describing both its current features and a number of ideas for improvement. We also report our experiences from a university course, where a group of computer science students worked like an expert community using OurWeb.

Challenges of urban sensing

Wireless sensor networks are emerging as a critical information technology, and they are continuing the trend originating in mainframe computing currently at the stage of mobile computing. This trend shows several aspects consistent in the evolution of computing including the increasing hardware miniaturization of the computing units and an increasing emphasis of the role of communication between the computing units – “networking”. In addition from the software side there is an increasing need to software solutions that are robust, exhibit distributed control, collaborative interfaces resulting in adaptive capabilities also at the system level. Like the present Internet, wireless sensor networks are large-scale distributed systems, but composed of smart sensors and actuators. They will eventually infuse the physical world and provide “grounding” for the Internet thus creating the Internet of Things. Research on wireless sensor networks has been taking place at several levels, from the lowest physical level to the highest information level – the latter is much less developed than the research at the physical levels. In addition, much of the research in wireless sensor networks has been focusing on military or science applications. However, wireless sensor networks can also play an important role in the realization of ubiquitous computing for everyday life – creating what we call “Urban sensing environment”. In urban sensing many natural gateways exist to collect and process the sensor information – static ones such as media devices, or mobile devices such as smart phones that can collect sensor information when entering the communication range of an active sensor. Some of the applications of wireless sensor network technology at home include, in addition to the surveillance functions, adding “intelligence” to utility consumption, electronic tagging, contamination control and disaster monitoring. Similarly at the community level “traffic monitoring” including people allows a development of totally unseen services from micro weather forecasts to new ways for “sensing the environment” for entertainment. In this talk we will outline some of the research challenges for urban sensing, and the role of learning and data analysis techniques for solving those challenges.

Evaluating the effect of social cues with automated experiments

Social cues present promising opportunities for facilitating collaboration online. Many interesting ideas and applications have been proposed, but there is still limited empirical understanding of the usefulness of various kinds of social cues. The contribution of this article is twofold. Firstly, we report a detailed evaluation of the social cues available in the collaborative learning environment EDUCO. Our study is based on automated experiments, in which the real social cues were occasionally replaced with artificial ones that were specifically designed for examining particular hypotheses. The results suggest that the users attend to the social cues, but do not make navigation decisions solely on the basis of social cues. Secondly, our study demonstrates the potential of automated experiments in studying behavior in virtual environments. Specific interventions made in natural circumstances can supplement other research methodologies in useful ways.ZusammenfassungDer Einsatz so genannter „social cues“ (soziale Anreize) stellt viel versprechende Möglichkelten zur Unterstützung von Online-Zusammenarbeit dar. Viele interessante Ideen und Anwendungen wurden diesbezüglich vorgestellt, aber empirisch belegte Erkenntnisse zur Nützlichkeit von social cues fehlen bis jetzt völlig. Dieser Artikel leistet zwei wesentliche Beiträge: Zunächst stellen die Autoren eine detaillierte Evaluation von social cues dar, basierend auf Erfahrungen in der Lernplattform EDUCO. Die Studie beruht auf automatisierten Experimenten, in denen reale social cues zufällig durch künstliche ersetzt wurden, um bestimmte Hypothesen zu testen. Die Ergebnisse zeigen, dass die End-Benutzer zwar soichen social cues Aufmerksamkeit schenken, aber ihre Navigationsentscheidungen davon nicht ausschließlich abhängig machen. Im Anschluss daran zeigt diese Studie das Potential automatisierter Experimente in Verhaltensstudien virtueller Räume.