Victor Codocedo

Bridging the gap between software architecture rationale formalisms and actual architecture documents: An ontology-driven approach

Documenting software architecture rationale is essential to reuse and evaluate architectures, and several modeling and documentation guidelines have been proposed in the literature. However, in practice creating and updating these documents rarely is a primary activity in most software projects, and rationale remains hidden in casual and semi-structured records, such as e-mails, meeting notes, wikis, and specialized documents. This paper describes the TREx (Toeska Rationale Extraction) approach to recover, represent and explore rationale information from text documents, combining: (1) pattern-based information extraction to recover rationale; (2) ontology-based representation of rationale and architectural concepts; and (3) facet-based interactive exploration of rationale. Initial results from TRExs application suggest that some kinds of architecture rationale can be semi-automatically extracted from a projects unstructured text documents, namely decisions, alternatives and requirements. The approach and some tools are illustrated with a case study of rationale recovery for a financial securities settlement system.

Pattern Structures and Concept Lattices for Data Mining and Knowledge Processing

This article aims at presenting recent advances in Formal Concept Analysis 2010-2015, especially when the question is dealing with complex data numbers, graphs, sequences, etc. in domains such as databases functional dependencies, data-mining local pattern discovery, information retrieval and information fusion. As these advances are mainly published in artificial intelligence and FCA dedicated venues, a dissemination towards data mining and machine learning is worthwhile.

Formal Concept Analysis and Information Retrieval – A Survey

One of the first models to be proposed as a document index for retrieval purposes was a lattice structure, decades before the introduction of Formal Concept Analysis. Nevertheless, the main notions that we consider so familiar within the community (“extension”, “intension”, “closure operators”, “order”) were already an important part of it. In the ’90s, as FCA was starting to settle as an epistemic community, lattice-based Information Retrieval (IR) systems smoothly transitioned towards FCA-based IR systems. Currently, FCA theory supports dozens of different retrieval applications, ranging from traditional document indices to file systems, recommendation, multi-media and more recently, semantic linked data. In this paper we present a comprehensive study on how FCA has been used to support IR systems. We try to be as exhaustive as possible by reviewing the last 25 years of research as chronicles of the domain, yet we are also concise in relating works by its theoretical foundations. We think that this survey can help future endeavours of establishing FCA as a valuable alternative for modern IR systems.

A Proposition for Combining Pattern Structures and Relational Concept Analysis

In this paper we propose an adaptation of the RCA process enabling the relational scaling of pattern structures. In a nutshell, this adaptation allows the scenario where RCA needs to be applied in a relational context family composed by pattern structures instead of formal contexts. To achieve this we define the heterogeneous pattern structures as a model to describe objects in a combination of spaces, namely the original object description space and the set of relational attributes derived from the RCA scaling process. We frame our approach in the problem of characterizing latent variables (LV) in a latent variable model of documents and terms. LVs are used as compact and improved dataset representations. We approach the problem of LV characterization missing from the original LV-model, through the application of the adapted RCA process using pattern structures. Finally, we discuss the implications of our proposition.

Lattice-based biclustering using partition pattern structures

In this work we present a novel technique for exhaustive bicluster enumeration using formal concept analysis (FCA). Particularly, we use pattern structures (an extension of FCA dealing with complex data) to mine similar row/column biclusters, a specialization of biclustering when attribute values have coherent variations. We show how biclustering can benefit from the FCA framework through its robust theoretical description and efficient algorithms. Finally, we evaluate our bicluster mining approach w.r.t. a standard biclustering technique showing very good results in terms of bicluster quality and performance.

When cyberathletes conceal their game: Clustering confusion matrices to identify avatar aliases

Video game is a very lucrative industry, unleashed by the ubiquity of gaming devices, multi-player networks and live broadcasting platforms. Games generate large amounts of behavioural data which are valuable to face the new challenges of video game analytics such as detecting balance issues, bugs and cheaters. In electronic sports (e-sports), cyberathletes conceal their online training using different aliases or avatars (virtual identities), which allow them not being recognized by the opponents they may face in future competitions (with cash prices challenging already most of the traditional sports). It was recently suggested that behavioural data generated by the games allows predicting the avatar associated to a game play with high accuracy. However, when a player uses several avatars, accuracy drastically drops as prediction models cannot easily differentiate the players different avatar aliases. Since mappings between players and avatars do not exist, we introduce the avatar aliases identification problem and propose an original approach for alias resolution based on supervised classification and Formal Concept Analysis. We thoroughly evaluate our method with the video game Starcraft 2 which has a very wide and active community with players from diverse cultures and nations. We show that under some circumstances, the avatars of a given player can easily be recognized as such. These results are valuable for e-sport structures (to help preparing tournaments), and game editors (detecting cheaters or usurpers).

No mining, no meaning: relating documents across repositories with ontology-driven information extraction

Far from eliminating documents as some expected, the Internet has lead to a proliferation of digital documents, without a centralized control or indexing. Thus, identifying relevant documents becomes simultaneously more important and much harder, since what users require may be dispersed across many documents and many repositories. This paper describes Ontologic Anchoring, a technique to relate documents in domain ontologies, using named entity recognition (a natural-language processing approach) and semantic annotation to relate individual documents to elements in ontologies. This approach allows document retrieval using domain-level inferences, and integration of repositories with heterogeneous media, languages and structure. Ontological anchoring is a two-way street: ontologies allow semantic indexing of documents, and simultaneously new documents enrich ontologies. The approach is illustrated with an initial deployment for heritage documents in Spanish.

Characterizing approximate-matching dependencies in formal concept analysis with pattern structures

Abstract Functional dependencies (FDs) provide valuable knowledge on the relations between attributes of a data table. A functional dependency holds when the values of an attribute can be determined by another. It has been shown that FDs can be expressed in terms of partitions of tuples that are in agreement w.r.t. the values taken by some subsets of attributes. To extend the use of FDs, several generalizations have been proposed. In this work, we study approximate-matching dependencies that generalize FDs by relaxing the constraints on the attributes, i.e. agreement is based on a similarity relation rather than on equality. Such dependencies are attracting attention in the database field since they allow uncrisping the basic notion of FDs extending its application to many different fields, such as data quality, data mining, behavior analysis, data cleaning or data partition, among others. We show that these dependencies can be formalized in the framework of Formal Concept Analysis (FCA) using a previous formalization introduced for standard FDs. Our new results state that, starting from the conceptual structure of a pattern structure, and generalizing the notion of relation between tuples, approximate-matching dependencies can be characterized as implications in a pattern concept lattice. We finally show how to use basic FCA algorithms to construct a pattern concept lattice that entails these dependencies after a slight and tractable binarization of the original data.

A Proposition for Sequence Mining Using Pattern Structures

In this article we present a novel approach to rare sequence mining using pattern structures. Particularly, we are interested in mining closed sequences, a type of maximal sub-element which allows providing a succinct description of the patterns in a sequence database. We present and describe a sequence pattern structure model in which rare closed subsequences can be easily encoded. We also propose a discussion and characterization of the search space of closed sequences and, through the notion of sequence alignments, provide an intuitive implementation of a similarity operator for the sequence pattern structure based on directed acyclic graphs. Finally, we provide an experimental evaluation of our approach in comparison with state-of-the-art closed sequence mining algorithms showing that our approach can largely outperform them when dealing with large regions of the search space.

Mining the Lattice of Binary Classifiers for Identifying Duplicate Labels in Behavioral Data

Analysis of behavioral data represents today a big issue, as so many domains generate huge quantity of activity and mobility traces. When traces are labeled by the user that generates it, models can be learned to accurately predict the user of an unknown trace. In online systems however, users may have several virtual identities, or duplicate labels. By ignoring them, the prediction accuracy drastically drops, as the set of all virtual identities of a single person is not known beforehand. In this article, we tackle this duplicate labels identification problem, and present an original approach that explores the lattice of binary classifiers. Each subset of labels is learned as the positive class against the others (the negative class), and constraints make possible to identify duplicate labels while pruning the search space. We experiment this original approach with data of the video game Starcraft 2 in the new context of Electronic Sports (eSport) with encouraging results.