Pablo Loyola

Predicting web user behavior using learning-based ant colony optimization

An ant colony optimization-based algorithm to predict web usage patterns is presented. Our methodology incorporates multiple data sources, such as web content and structure, as well as web usage. The model is based on a continuous learning strategy based on previous usage in which artificial ants try to fit their sessions with real usage through the modification of a text preference vector. Subsequently, trained ants are released onto a new web graph and the new artificial sessions are compared with real sessions, previously captured via web log processing. The main results of this work are related to an effective prediction of the aggregated patterns of real usage, reaching approximately 80%. In the second place, this approach allows the obtaining of a quantitative representation of the keywords that influence the navigational sessions.

Dodona: automated oracle data set selection

Software complexity has increased the need for automated software testing. Most research on automating testing, however, has focused on creating test input data. While careful selection of input data is necessary to reach faulty states in a system under test, test oracles are needed to actually detect failures. In this work, we describe Dodona, a system that supports the generation of test oracles. Dodona ranks program variables based on the interactions and dependencies observed between them during program execution. Using this ranking, Dodona proposes a set of variables to be monitored, that can be used by engineers to construct assertion-based oracles. Our empirical study of Dodona reveals that it is more effective and efficient than the current state-of-the-art approach for generating oracle data sets, and can often yield oracles that are almost as effective as oracles hand-crafted by engineers without support.

A Neural Architecture for Generating Natural Language Descriptions from Source Code Changes.

We propose a model to automatically describe changes introduced in the source code of a program using natural language. Our method receives as input a set of code commits, which contains both the modifications and message introduced by an user. These two modalities are used to train an encoder-decoder architecture. We evaluated our approach on twelve real world open source projects from four different programming languages. Quantitative and qualitative results showed that the proposed approach can generate feasible and semantically sound descriptions not only in standard in-project settings, but also in a cross-project setting.

Clustering-Based Learning Approach for Ant Colony Optimization Model to Simulate Web User Behavior

In this paper we propose a novel methodology for analyzing web user behavior based on session simulation by using an Ant Colony Optimization algorithm which incorporates usage, structure and content data originating from a real web site. In the first place, artificial ants learn from a clustered web user session set through the modification of a text preference vector. Then, trained ants are released through a web graph and the generated artificial sessions are compared with real usage. The main result is that the proposed model explains approximately 80% of real usage in terms of a predefined similarity measure.

Modeling User Session and Intent with an Attention-based Encoder-Decoder Architecture

We propose an encoder-decoder neural architecture to model user session and intent using browsing and purchasing data from a large e-commerce company. We begin by identifying the source-target transition pairs between items within each session. Then, the set of source items are passed through an encoder, whose learned representation is used by the decoder to estimate the sequence of target items. Therefore, as this process is performed pair-wise, we hypothesize that the model could capture the transition regularities in a more fine grained way. Additionally, our model incorporates an attention mechanism to explicitly learn the more expressive portions of the sequences in order to improve performance. Besides modeling the user sessions, we also extended the original architecture by means of attaching a second decoder that is jointly trained to predict the purchasing intent of user in each session. With this, we want to explore to what extent the model can capture inter session dependencies. We performed an empirical study comparing against several baselines on a large real world dataset, showing that our approach is competitive in both item and intent prediction.

Characterizing Web User Visual Gaze Patterns: A Graph Theory Inspired Approach

We propose a graph-based analysis framework to study the dynamics of visual gaze from web users. Our goal is to extract the main characteristics of the information foraging process from an attention-centric perspective. Our approach consists of modeling web objects, such as images and paragraphs, as nodes. The visual transitions are represented as edges. With the resulting graphs, several standard metrics were computed. We performed an initial empirical study with 23 subjects. The visual activity was captured using an eye tracking device. The results suggest that a graph based analysis can capture in a reliable way the dynamics of user behavior and the identification of salient objects within a web site.

Population dynamics in open source communities: an ecological approach applied to github

Open Source Software (OSS) has gained high amount of popularity during the last few years. It is becoming used by public and private institutions, even companies release portions of their code to obtain feedback from the community of voluntary developers. As OSS is based on the voluntary contributions of developers, the number of participants represents one of the key elements that impact the quality of the software. In order to understand how the the population of contributors evolve over time, we propose a methodology that adapts Lotka-Volterra-based biological models used for describing host-parasite interactions. Experiments based on data from the Github collaborative platform showed that the proposed approach performs effectively in terms of providing an estimation of the population of developers for each project over time.

Identifying user sessions from web server logs with integer programming

Web usage mining has proven to be an important advance for e-business systems, both by finding web user buying patterns and suggesting ways to improve web user navigation. A primary input for web usage mining is web user sessions that must be constructed from web server logs called sessionization when such sessions are not otherwise identified. We use bipartite cardinality matching and a more general integer program to construct sessions. We also propose several variations of our integer program to provide additional insights into session characteristics. For testing, we retrieve 15 months of web server logs and corresponding real sessions from an academic web site. We compare real sessions, results obtained by our optimization models, and results from a commonly-used timeout heuristic. We find our optimization models dominate the timeout heuristic using several comparison measures. Solution time for a typical month is seven hours for our integer program, 30 minutes for our bipartite cardinality matching, and about 1 minute for the heuristic. Although solution time is significantly greater for the integer program, its variations contribute additional analysis of web user behavior.

Biological Mutualistic Models Applied to Study Open Source Software Development

The evolution of the Web has allowed the generation of several platforms for collaborative work. One of the main contributors to these advances is the Open Source initiative, in which projects are boosted to a new level of interaction and cooperation that improves their software quality and reliability. In order to understand how the group of contributors interacts with the software under development, we propose a novel methodology that adapts Lotka-Volterra-based biological models used for host-parasite interaction. In that sense, we used the concept mutualism from social parasites. Preliminary results based on experiments on the Github collaborative platform showed that Open Source phenomena can be modeled as a mutualistic system, in terms of the evolution of the population of developers and repositories.

Learning graph representations for defect prediction

We propose to study the impact of the representation of the data in defect prediction models. For this study, we focus on the use of developer activity data, from which we structure dependency graphs. Then, instead of manually generating features, such as network metrics, we propose a model inspired in recent advances in Representation Learning which are able to automatically learn representations from graph data. These new representations are compared against manually crafted features for defect prediction in real world software projects.