Online social networks play a major role in the spread of information at very large scale and it becomes essential to provide means to analyze this phenomenon. Analyzing information diffusion proves to be a challenging task since the raw data produced by users of these networks are a flood of ideas, recommendations, opinions, etc. The aim of this PhD work is to help in the understanding of this phenomenon. So far, our contributions are the following: (i) a survey of developments in the field; (ii) T-BaSIC, a graph-based model for information diffusion prediction; (iii) SONDY, an open source platform that helps understanding social network users' interests and activity by providing emerging topics and events detection as well as network analysis functionalities.

In induction graphs methods such as C4.51 or SIPINA2, taking continuous attributes into account needs particular discretization procedures. In this paper, we propose on the one hand, an axiomatic leading to a set of criteria which can be used for continuous attributes discretization, and on the other hand, a method of discretization called FUSINTER. The results obtained by FUSINTER are compared to those obtained by techniques developed by Fayyad and Irani3 and Kerber4 and they have proved better for the majority of the examples studied.

We propose a new statistical approach for characterizing the class separability degree in ℝp. This approach is based on a non‐parametric statistic called ‘the cut edge weight’. We show in this paper the principle and the experimental applications of this statistic. First, we build a geometrical connected graph like Toussaint's Relative Neighbourhood Graph on all examples of the learning set. Second, we cut all edges between two examples of a different class. Third, we compute the relative weight of these cut edges. If the relative weight of the cut edges is in the expected range of a random distribution of the labels on all the neighbourhood of the graph's vertices, then no neighbourhood‐based method provides a reliable prediction model. We will say then that the classes to predict are non‐separable. Copyright © 2005 John Wiley & Sons, Ltd.

No abstract

Background and objective: Automated healthcare databases (AHDB) are an important data source for real life drug and healthcare use. In the filed of depression, lack of detailed clinical data requires the use of binary proxies with important limitations. The study objective was to create a Depressive Health State Index (DHSI) as a continuous health state measure for depressed patients using available data in an AHDB. Methods: The study was based on historical cohort design using the UK Clinical Practice Research Datalink (CPRD). Depressive episodes (depression diagnosis with an antidepressant prescription) were used to create the DHSI through 6 successive steps: (1) Defining study design; (2) Identifying constituent parameters; (3) Assigning relative weights to the parameters; (4) Ranking based on the presence of parameters; (5) Standardizing the rank of the DHSI; (6) Developing a regression model to derive the DHSI in any other sample. Results: The DHSI ranged from 0 (worst) to 100 (best health state) comprising 29 parameters. The proportion of depressive episodes with a remission proxy increased with DHSI quartiles. Conclusion: A continuous outcome for depressed patients treated by antidepressants was created in an AHDB using several different variables and allowed more granularity than currently used proxies.