François Trousset

Web opinion mining: how to extract opinions from blogs?

The growing popularity of Web 2.0 provides with increasing numbers of documents expressing opinions on different topics. Recently, new research approaches have been defined in order to automatically extract such opinions from the Internet. They usually consider opinions to be expressed through adjectives, and make extensive use of either general dictionaries or experts to provide the relevant adjectives. Unfortunately, these approaches suffer from the following drawback: in a specific domain, a given adjective may either not exist or have a different meaning from another domain. In this paper, we propose a new approach focusing on two steps. First, we automatically extract a learning dataset for a specific domain from the Internet. Secondly, from this learning set we extract the set of positive and negative adjectives relevant to the domain. The usefulness of our approach was demonstrated by experiments performed on real data.

Towards an automatic characterization of criteria

The number of documents is growing exponentially with the rapid expansion of the Web. The new challenge for Internet users is now to rapidly find appropriate data to their requests. Thus information retrieval, automatic classification and detection of opinions appear as major issues in our information society. Many efficient tools have already been proposed to Internet users to ease their search over the web and support them in their choices. Nowadays, users would like genuine decision tools that would efficiently support them when focusing on relevant information according to specific criteria in their area of interest. In this paper, we propose a new approach for automatic characterization of such criteria. We bring out that this approach is able to automatically build a relevant lexicon for each criterion. We then show how this lexicon can be useful for documents classification or segmentation tasks. Experiments have been carried out with real datasets and show the efficiency of our proposal.

Privacy preserving sequential pattern mining in distributed databases

Research in the areas of privacy preserving techniques in databases and subsequently in privacy enhancement technologies have witnessed an explosive growth-spurt in recent years. This escalation has been fueled by the growing mistrust of individuals towards organizations collecting and disbursing their Personally Identifiable Information (PII). Digital repositories have become increasingly susceptible to intentional or unintentional abuse, resulting in organizations to be liable under the privacy legislations that are being adopted by governments the world over. These privacy concerns have necessitated new advancements in the field of distributed data mining wherein, collaborating parties may be legally bound not to reveal the private information of their customers. In this paper, we present a new algorithm PriPSeP (Privacy Preserving SEquential Patterns) for the mining of sequential patterns from distributed databases while preserving privacy. A salient feature of PriPSeP is that due to its flexibility it is more pertinent to mining operations for real world applications in terms of efficiency and functionality. Under some reasonable assumptions, we prove that our architecture and protocol employed by our algorithm for multi-party computation is secure.

Opinion Extraction Applied to Criteria

The success of Information technologies and associated services (e.g., blogs, forums,...) eases the way to express massive opinion on various topics. Recently new techniques known as opinion mining have emerged. One of their main goals is to automatically extract a global trend from expressed opinions. While it is quite easy to get this overall assessment, a more detailed analysis will highlight that opinions are expressed on more specific topics: one will acclaim a movie for its soundtrack and another will criticize it for its scenario. Opinion mining approaches have little explored this multicriteria aspect. In this paper we propose an automatic extraction of text segments related to a set of criteria. The opinion expressed in each text segment is then automatically extracted. From a small set of opinion keywords, our approach automatically builds a training set of texts from the web. A lexicon reflecting the polarity of words is then extracted from this training corpus. This lexicon is then used to compute the polarity of extracted text segments. Experiments show the efficiency of our approach.

Multi-criteria improvement of options

Designing the way a complex system should evolve to better match the customers’ requirements provides an interesting class of applications for muticriteria techniques. The required models to support the improvement design of a complex system must include both preference models and system behavioral models. A MAUT model captures the decisions related to customers’ preferences whereas a fuzzy representation is proposed to model the relationships between systems parameters and performances to capture operational constraints. This latter part of the improvement design is supported by a branch and bound algorithm to efficiently compute the most relevant actions to be performed.

On the consideration of a bring-to-mind model for computing the Information Content of concepts defined into ontologies

Ontologies are core elements of numerous applications that are based on computer-processable expert knowledge. They can be used to estimate the Information Content (IC) of the key concepts of a domain: a central notion on which depend various ontology-driven analyses, e.g. semantic measures. This paper proposes new IC models based on the belief functions theoretical framework. These models overcome limitations of existing ICs that do not consider the inductive inference assumption intuitively assumed by human operators, i.e. that occurrences of a concept (e.g. Maths) not only impact the IC of more general concepts (e.g. Sciences), as considered by traditional IC models, but also the one of more specific concepts (e.g. Algebra). Interestingly, empirical evaluations show that, in addition to modelling the aforementioned assumption, proposed IC models compete with best state-of-the-art models in several evaluation settings.

Comfort as a multidimensional preference model for energy efficiency control issues

The incessant need for energy has raised its cost to unexpected heights. In response to this situation, many projects have been started in order to save energy. In this context, RIDER project tries to develop a weak system dependency of energy management framework which could be applied for different systems. Particularly, our RIDER Decision Support System (DSS) focuses on proposing generic control rules and optimization techniques for energy management systems. Therefore, the DSS aims to compute the most relevant target values (i.e., setpoints) to be provided to the energy control system and then, improving thermal comfort sensation or reducing energy costs. Literature proposes reusable system independent statistical models for thermal comfort. However, they are not easily interpretable in terms of a preference model which makes control not intuitive and tractable. Since thermal comfort is a subjective multidimensional concept, an interpretable and reusable preference model is introduced in this paper. Multi Attribute Utility Theory (MAUT) is used for this.

The Translation of Will into Act: Achieving a Consensus between Managerial Decisions and Operational Abilities

Abstract In the current industrial context, strategies intended to bring about continuous improvement have to include the multi-criteria performance expression aspects. A MAUT model is proposed in the first part of this paper: it captures the managers’ strategy in terms of performances improvement. The search of an efficient improvement is formalized as an optimization problem. Nevertheless, MAUT models address purely managerial decisions but do not include the material constraints related to the action plans that address the required improvement. A qualitative model is thus proposed to support this implementation part. It models the relations between goals and actions to define the most relevant action plan. Finally, a unified framework is proposed to conciliate managerial and implementation aspects in an industrial improvement project. It integrates a preferences’ model for the managerial aspects and a CSP model for the operational aspects. Both models are conjointly run into an iterative process to define an efficient improvement.

Intrusion Detections in Collaborative Organizations by Preserving Privacy

To overcome the problem of attacks on networks, new Intrusion Detection System (IDS) approaches have been proposed in recent years. They consist in identifying signatures of known attacks to compare them to each request and determine whether it is an attack or not. However, these methods are set to default when the attack is unknown from the database of signatures. Usually this problem is solved by calling human expertise to update the database of signatures. However, it is frequent that an attack has already been detected by another organization and it would be useful to be able to benefit from this knowledge to enrich the database of signatures. Unfortunately this information is not so easy to obtain. In fact organizations do not necessarily want to spread the information that they have already faced this type of attack. In this paper we propose a new approach to intrusion detection in a collaborative environment but by preserving the privacy of the collaborative organizations. Our approach works for any signature that may be written as a regular expression insuring that no information is disclosed on the content of the sites.

SAX: a privacy preserving general pupose methodapplied to detection of intrusions

To overcome the problem of attacks on networks, new Intrusion Detection System (IDS) approaches have been proposed in recent years. They consist in identifying signatures of known attacks to compare them to each request and determine whether it is an attack or not. However, these methods are set to default when the attack is unknown from the database of signatures. Usually this problem is solved by calling human expertise to update the database of signatures. However, it is frequent that an attack has already been detected by another organization and it would be useful to be able to benefit from this knowledge to enrich the database of signatures. Unfortunately this information is not so easy to obtain. In fact organizations do not necessarily want to spread the information that they have already faced this type of attack. In this paper we propose a new approach to intrusion detection in a collaborative environment but by preserving the privacy of the collaborative organizations. Our approach works for any signature even if it needs a complex program to be detected and insure that no information is disclosed on the content of any of the sites. For this pupose, we have developped a general method (SAX) that allows to compute any algorithm while preserving privacy of data and also of the program code which is computed.