Andrea Addis

A recommender system based on a generic contextual advertising approach

Contextual advertising and automatic recommendation are emerging fields deeply studied by researchers in information retrieval. So far, these fields have been separately investigated and several solutions have been independently proposed in the literature. Nevertheless, in our view, there are common issues that drive us to think that systems devised to solve one of the task above could be simply re-adapted to solve the other. In this paper, we propose a novel recommendation system based on a generic solution typically adopted to solve contextual advertising tasks. Experimental results highlight the effectiveness of the proposed approach.

From Unstructured Web Knowledge to Plan Descriptions

Automated Planning (AP) is an AI field whose goal is to automatically generate sequence of actions that solve problems. One of the main difficulties in its extensive use in real-world application lies in the fact that it requires the careful and error-prone process of defining a declarative domain model. This is usually performed by planning experts who should know about both the domain in hand, and the planning techniques (including sometimes the inners of these techniques or the tools that implement them). In order planning to be widely used, this process should be performed by non-planning experts. On the other hand, in many domains there are plenty of electronic documents (including the Web) that describe processes or plans in a semi-structured way. These descriptions mix natural language and certain templates for that specific domain. One such examples is the www.WikiHow.com web site that includes plans in many domains, all plans described through a set of common templates. In this work, we present a suite of tools that automatically extract knowledge from those unstructured descriptions of plans to be used for diverse planning applications.

A comparative experimental assessment of a threshold selection algorithm in hierarchical text categorization

Most of the research on text categorization has focused on mapping text documents to a set of categories among which structural relationships hold, i.e., on hierarchical text categorization. For solutions of a hierarchical problem that make use of an ensemble of classifiers, the behavior of each classifier typically depends on an acceptance threshold, which turns a degree of membership into a dichotomous decision. In principle, the problem of finding the best acceptance thresholds for a set of classifiers related with taxonomic relationships is a hard problem. Hence, devising effective ways for finding suboptimal solutions to this problem may have great importance. In this paper, we assess a greedy threshold selection algorithm aimed at finding a suboptimal combination of thresholds in a hierarchical text categorization setting. Comparative experiments, performed on Reuters, report the performance of the proposed threshold selection algorithm against a relaxed brute-force algorithm and against two state-of-the-art algorithms. Results highlight the effectiveness of the approach.

Using Progressive Filtering to Deal with Information Overload

In the age of Web 2.0 people organize large collections of web pages, articles, or emails in hierarchies of topics, or arrange a large body of knowledge in ontologies. This scenario requires automatic text categorization systems able to cope with underlying taxonomies in an effective and efficient way, so that information overload and input imbalance can be suitably dealt with. In this work, we propose a hierarchical text categorization approach that decomposes a given rooted taxonomy into pipelines, one for each path that exists between the root and each node of the taxonomy, so that each pipeline can be tuned in isolation. Experimental results, performed on Reuters and DMOZ data collections, show that the proposed approach performs better than a flat approach in presence of input imbalance.

Multiagent systems and information retrieval our experience with X.MAS

Highlights? A multiagent architecture for implementing information retrieval and filtering tasks. ? Personalization issues in information retrieval. ? Six applications that demonstrate effectiveness and usability of the architecture. ? Lesson learnt in applying multiagent systems in the field of information retrieval. Autonomous agents and multiagent systems have been successfully applied to a number of problems and have been largely used in different application fields. In particular, in this paper we are interested in information retrieval. In fact, in this field multiagent solutions are very useful and effective since they decouple the problem in a network of software agents that interact to solve problems that are beyond the individual capabilities or knowledge. In so doing, multiagent systems allow to overwhelm typical problems of single agent and centralized approaches. To discuss the lesson learnt in using the multiagent technology in the field of information retrieval, in this paper, we present our experience in using X.MAS, a generic multiagent architecture aimed at retrieving, filtering and reorganizing information according to user interests. To this end, after presenting X.MAS, we illustrate six applications built upon it. Our conclusion is that multiagent technology is quite effective to design and realize concrete information retrieval applications.

Experimentally Studying Progressive Filtering in Presence of Input Imbalance

Progressively Filtering (PF) is a simple categorization technique framed within the local classifier per node approach. In PF, each classifier is entrusted with deciding whether the input in hand can be forwarded or not to its children. A simple way to implement PF consists of unfolding the given taxonomy into pipelines of classifiers. In so doing, each node of the pipeline is a binary classifier able to recognize whether or not an input belongs to the corresponding class. In this chapter, we illustrate and discuss the results obtained by assessing the PF technique, used to perform text categorization. Experiments, on the Reuters Corpus (RCV1- v2) dataset, are focused on the ability of PF to deal with input imbalance. In particular, the baseline is: (i) comparing the results to those calculated resorting to the corresponding flat approach; (ii) calculating the improvement of performance while augmenting the pipeline depth; and (iii) measuring the performance in terms of generalization- / specialization- / misclassification-error and unknown-ratio. Experimental results show that, for the adopted dataset, PF is able to counteract great imbalances between negative and positive examples. We also present and discuss further experiments aimed at assessing TSA, the greedy threshold selection algorithm adopted to perform PF, against a relaxed brute-force algorithm and the most relevant state-of-the-art algorithms.

Hierarchical Text Categorization Through a Vertical Composition of Classifiers

In this paper we present a hierarchical approach to text categorization aimed at improving the performances of the corresponding tasks. The proposed approach is explicitly devoted to cope with the problem related to the unbalance between relevant and non relevant inputs. The technique has been implemented and tested by resorting to a multiagent system aimed at performing information retrieval tasks.

A Comparative Study of Thresholding Strategies in Progressive Filtering

Thresholding strategies in automated text categorization are an underexplored area of research. Indeed, thresholding strategies are often considered a post-processing step of minor importance, the underlying assumptions being that they do not make a difference in the performance of a classifier and that finding the optimal thresholding strategy for any given classifier is trivial. Neither these assumptions are true. In this paper, we concentrate on progressive filtering, a hierarchical text categorization technique that relies on a local-classifier-per-node approach, thus mimicking the underlying taxonomy of categories. The focus of the paper is on assessing TSA, a greedy threshold selection algorithm, against a relaxed brute-force algorithm and the most relevant state-of-the-art algorithms. Experiments, performed on Reuters, confirm the validity of TSA.

Progressive Filtering on the Web: The Press Reviews Case Study

Progressive Filtering is a hierarchical classification technique framed within the local classifier per node approach where each classifier is entrusted with deciding whether the input in hand can be forwarded or not to its children. In this chapter, we illustrate the effectiveness of Progressive Filtering on the Web, focusing on the task of automatically creating press reviews. To this end, we present NEWS.MAS, a multiagent system aimed at: (i) extracting information from online newspapers by using suitable wrapper agents, each associated with a specific information source, (ii) categorizing news articles according to a given taxonomy, and (iii) providing user feedback to improve the performance of the system depending on user needs and preferences.