Carlos Castro-Herrera

On-demand feature recommendations derived from mining public product descriptions

We present a recommender system that models and recommends product features for a given domain. Our approach mines product descriptions from publicly available online specifications, utilizes text mining and a novel incremental diffusive clustering algorithm to discover domain-specific features, generates a probabilistic feature model that represents commonalities, variants, and cross-category features, and then uses association rule mining and the k-Nearest-Neighbor machine learning strategy to generate product specific feature recommendations. Our recommender system supports the relatively labor-intensive task of domain analysis, potentially increasing opportunities for re-use, reducing time-to-market, and delivering more competitive software products. The approach is empirically validated against 20 different product categories using thousands of product descriptions mined from a repository of free software applications.

Supporting Domain Analysis through Mining and Recommending Features from Online Product Listings

Domain analysis is a labor-intensive task in which related software systems are analyzed to discover their common and variable parts. Many software projects include extensive domain analysis activities, intended to jumpstart the requirements process through identifying potential features. In this paper, we present a recommender system that is designed to reduce the human effort of performing domain analysis. Our approach relies on data mining techniques to discover common features across products as well as relationships among those features. We use a novel incremental diffusive algorithm to extract features from online product descriptions, and then employ association rule mining and the (k)-nearest neighbor machine learning method to make feature recommendations during the domain analysis process. Our feature mining and feature recommendation algorithms are quantitatively evaluated and the results are presented. Also, the performance of the recommender system is illustrated and evaluated within the context of a case study for an enterprise-level collaborative software suite. The results clearly highlight the benefits and limitations of our approach, as well as the necessary preconditions for its success.

Automated support for managing feature requests in open forums

The result is stable, focused, dynamic discussion threads that avoid redundant ideas and engage thousands of stakeholders.

Using Data Mining and Recommender Systems to Facilitate Large-Scale, Open, and Inclusive Requirements Elicitation Processes

Requirements related problems, especially those originating from inadequacies in the human-intensive task of eliciting stakeholderspsila needs and desires, have contributed to many failed and challenged software projects. This is especially true for large and complex projects in which requirements knowledge is distributed across thousands of stakeholders. This short paper introduces a new process and related framework that utilizes data mining and recommender technologies to create an open, scalable, and inclusive requirements elicitation process capable of supporting projects with thousands of stakeholders. The approach is illustrated and evaluated using feature requests mined from an open source software product.

Enhancing Stakeholder Profiles to Improve Recommendations in Online Requirements Elicitation

Requirements elicitation has long been recognized as a crucial activity in any software development project. Unfortunately, the traditional elicitation practices do not scale well when applied to larger projects, where knowledge is distributed across numerous geographically dispersed stakeholders. As a result, new distributed requirements elicitation tools have started to surface, such as online forums and wiki pages. In our previous work, we introduced a framework for supporting distributed elicitation by utilizing data mining and machine learning techniques to automatically group stakeholder ideas into forums, and by using recommender system technologies to help promote these forums to potentially interested stakeholders. The framework is designed to create an open and more inclusive environment where points of view, conflicts, interests and tradeoffs are identified as early as possible. In this paper, we present two substantial enhancements to the Recommender System component of this framework, and demonstrate through experiments how they improve the quality of the recommendations.

Utilizing recommender systems to support software requirements elicitation

Requirements Engineering involves a number of human intensive activities designed to help project stakeholders discover, analyze, and specify the functional and non-functional needs for a software intensive system. Recommender systems can support several different areas of this process including identifying potential subject matter experts for a topic, keeping individual stakeholders informed of relevant issues, and even recommending possible features for stakeholders to consider and explore. This position paper summarizes an extensive series of experiments that were conducted to identify best-of-breed algorithms for recommending forums to stakeholders and recommending unexplored topics to project managers.

A recommender system for dynamically evolving online forums

Recommender systems can be used in online forums to recommend discussion topics to users; however as these forums are characterized by a constant influx of new users and new posts, it is important to consider the performance of the recommender system under a scenario in which the internal composition of the items to be recommended, i.e., discussion threads, and the user preferences are constantly changing. In this paper we describe and evaluate a forum recommender designed to handle the challenges of dynamically evolving internet forums used to gather and discuss feature requests for various software products. In particular, we empirically show that two proposed enhancements to the representations of user profiles will result in improved recommendation effectiveness in dynamic environments.

Recommendation Systems in Requirements Discovery

Recommendation systems offer the opportunity for supporting and enhancing a wide variety of activities in requirements engineering. We discuss several potential uses. In particular we highlight the role of recommendation systems in online forums that are used for capturing and discussing feature requests. The recommendation system is used to mitigate problems introduced when face-to-face communication is replaced with potentially high-volume online discussions. In this context, recommendation systems can be used to suggest relevant topics to stakeholders and conversely to recommend expert stakeholders for each discussion topic. We also explore the use of recommendation systems in the domain analysis process, where they can be used to recommend sets of features to include in new products.

A genetic algorithm based framework for software effort prediction

BackgroundSeveral prediction models have been proposed in the literature using different techniques obtaining different results in different contexts. The need for accurate effort predictions for projects is one of the most critical and complex issues in the software industry. The automated selection and the combination of techniques in alternative ways could improve the overall accuracy of the prediction models.ObjectivesIn this study, we validate an automated genetic framework, and then conduct a sensitivity analysis across different genetic configurations. Following is the comparison of the framework with a baseline random guessing and an exhaustive framework. Lastly, we investigate the performance results of the best learning schemes.MethodsIn total, six hundred learning schemes that include the combination of eight data preprocessors, five attribute selectors and fifteen modeling techniques represent our search space. The genetic framework, through the elitism technique, selects the best learning schemes automatically. The best learning scheme in this context means the combination of data preprocessing + attribute selection + learning algorithm with the highest coefficient correlation possible. The selected learning schemes are applied to eight datasets extracted from the ISBSG R12 Dataset.ResultsThe genetic framework performs as good as an exhaustive framework. The analysis of the standardized accuracy (SA) measure revealed that all best learning schemes selected by the genetic framework outperforms the baseline random guessing by 45–80%. The sensitivity analysis confirms the stability between different genetic configurations.ConclusionsThe genetic framework is stable, performs better than a random guessing approach, and is as good as an exhaustive framework. Our results confirm previous ones in the field, simple regression techniques with transformations could perform as well as nonlinear techniques, and ensembles of learning machines techniques such as SMO, M5P or M5R could optimize effort predictions.