CCOMPUT - ArtículosCCOMPUT - Artículoshttp://hdl.handle.net/10017/3042024-03-29T14:19:15Z2024-03-29T14:19:15ZUnsupervised intrusion detection through skip-gram models of network behaviorSan Miguel Carrasco, RafaelSicilia Urbán, Miguel Ángelhttp://hdl.handle.net/10017/610752024-03-15T09:36:01Z2018-07-12T00:00:00ZUnsupervised intrusion detection through skip-gram models of network behavior
San Miguel Carrasco, Rafael; Sicilia Urbán, Miguel Ángel
Detecting intrusions is one of the main objectives of computer security. Attacks have become overly sophisticated over the years in order to remain effective and stealthy. Major breaches are typically perpetrated using techniques that are polymorphic, multi-vector, multi-stage and targeted, that is, adopting forms that were never seen before. Anomaly detection, which does not make any assumption about the shape of a potential attack but instead on legitimate behavior, seems to be a suitable approach in order to defeat sophisticated intrusions. Skip-gram modeling, a word2vec algorithm variant, was leveraged to model systems" legitimate network behavior. The resulting model was then used to spot intrusions in a test dataset. The optimal configuration led to 99.20% precision, 82.07% recall, and 91.02% accuracy, with a false positive rate of 0.61%, which is significantly lower than most state-of-the-art methods. These metrics were achieved under a fully unsupervised setting, that is, without any prior knowledge of what constitutes an attack. Furthermore, the approach provides benefits in terms of interpretability and log storage requirements, as it requires a small amount of input features. It also produces information about systems behavior and their relationships, that can be reused by other analysis techniques to obtain further insights.
2018-07-12T00:00:00ZTeam efficiency and network structure: The case of professional League of LegendsMora Cantallops, MarçalSicilia Urbán, Miguel Ángelhttp://hdl.handle.net/10017/610742024-03-18T07:50:44Z2019-04-03T00:00:00ZTeam efficiency and network structure: The case of professional League of Legends
Mora Cantallops, Marçal; Sicilia Urbán, Miguel Ángel
Teams can be defined by their interactions and successful performance rests on their members"
behaviour. Although this topic has been studied both in sports and management, research
on computer mediated team interactions, communication, cooperative work and efficiency
in online competitive environments is scarce. In this article, networks will be used as a
novel approach to understand how League of Legends professional players assist each other
during a competitive match and to link their computer mediated behaviour and social
interactions to their team"s performance. Starting from a dataset consisting of 453.386
kill assists, the network structure and efficiency is assessed over 7.582 matches in total.
After controlling for potential mixed-effects, such as the quality of the involved teams or
their geography, this study reinforces previous research showing that team efficiency in the
League of Legends professional scene is positively affected by the intensity of their interaction
while centralization of resources is detrimental. Networks with high intensity and low inner
centralization are, therefore, related to a higher performance as a team not only in traditional
sports but also in computer mediated contexts.
2019-04-03T00:00:00ZPlayer-centric networks in League of LegendsMora Cantallops, MarçalSicilia Urbán, Miguel Ángelhttp://hdl.handle.net/10017/610732024-03-20T11:39:14Z2018-06-27T00:00:00ZPlayer-centric networks in League of Legends
Mora Cantallops, Marçal; Sicilia Urbán, Miguel Ángel
Online competitive gaming has become one of the largest collective human activities globally and understanding motivations and social interaction is still not fully achieved. The aim of this study is to develop a basis for a systematic classification of player-centric networks in competitive online games based on structural network criteria. Using data extracted from League of Legends players, their matches and machine learning techniques, a classification of personal player networks in League of Legends is proposed. Results show the resulting egonets can be potentially grouped in four clusters related to their egos playing habits, ranging from solo to team players.
2018-06-27T00:00:00ZAutomated creation of an intent model for conversational agentsBenayas, AlbertoSicilia Urbán, Miguel ÁngelMora Cantallops, Marçalhttp://hdl.handle.net/10017/610352024-03-12T01:17:59Z2023-01-20T00:00:00ZAutomated creation of an intent model for conversational agents
Benayas, Alberto; Sicilia Urbán, Miguel Ángel; Mora Cantallops, Marçal
Conversational Agents (CA) are increasingly being deployed by organizations to provide round-the-clock support and to increase customer satisfaction. All CA have one thing in common despite the differences in their design: they need to be trained with users' intents and corresponding training sentences. Access to proper data with acceptable coverage of intents and training sentences is a big challenge in CA deployment. Even with the access to the past conversations, the process of discovering intents and training sentences manually is not time and cost-effective. Here, an end to end automated framework that can discover intents and their training sentences in conversation logs to generate labeled data sets for training intent models is presented. The framework proposes different feature engineering techniques and leverages dimensionality reduction methods to assemble the features, then applies a density-based clustering algorithm iteratively to mine even the least common intents. Finally, the clustering results are automatically labeled by the final algorithm.
2023-01-20T00:00:00Z