Marcelo G. Armentano

Topology-Based Recommendation of Users in Micro-Blogging Communities

Nowadays, more and more users share real-time news and information in micro-blogging communities such as Twitter, Tumblr or Plurk. In these sites, information is shared via a followers/followees social network structure in which a follower will receive all the micro-blogs from the users he/she follows, named followees. With the increasing number of registered users in this kind of sites, finding relevant and reliable sources of information becomes essential. The reduced number of characters present in micro-posts along with the informal language commonly used in these sites make it difficult to apply standard content-based approaches to the problem of user recommendation. To address this problem, we propose an algorithm for recommending relevant users that explores the topology of the network considering different factors that allow us to identify users that can be considered good information sources. Experimental evaluation conducted with a group of users is reported, demonstrating the potential of the approach.

Plan recognition for interface agents

Interface agents are computer programs that provide personalized assistance to a user dealing with computer based applications. By understanding the tasks the user performs in a software application an interface agent could be aware of the context that represents the user’s focus of attention at each particular moment. With this purpose, plan recognition aims at identifying the plans or goals of a user from the tasks he (for simplicity, we use “he” to refer to the user, but we do not mean any distinctions about sexes) performs. A prerequisite for the recognition of plans is knowledge of a user’s possible tasks and the combination of these tasks in complex task sequences, which describes typical user behavior. Plan recognition will enable an interface agent to reason about what the user might do next so that it can determine how to assist him. In this work we present the state of the art in Plan Recognition, paying special attention to the features that make it useful to interface agents. These features include the ability to deal with uncertainty, multiple plans, multiple interleaved goals, overloaded tasks, noisy tasks, interruptions and the capability to adapt to a particular user.

Recognition of User Intentions for Interface Agents with Variable Order Markov Models

A key aspect to study in the field of interface agents is the need to detect as soon as possible the user intentions. User intentions have an important role for an interface agent because they serve as a context to define the way in which the agents can collaborate with users. Intention recognition can be used to infer the users intentions based on the observation of the tasks the user performs in a software application. In this work, we propose an approach to model the intentions the user can pursue in an application in a semi-automatic way, based on Variable-Order Markov models. We claim that with appropriate training from the user, an interface agent following our approach will be able both to detect the user intention and the most probable sequence of following tasks the user will perform to achieve his/her intention.

Personal assistants: Direct manipulation vs. mixed initiative interfaces

Interface agents are computer programmes that provide assistance to users dealing with computer-based applications. The introduction of agents to user interfaces caused the exploration of new metaphors to enhance user ability to directly manipulate interfaces. In this regard, mixed-initiative interaction refers to a flexible interaction strategy in which agents contribute with users by providing suitable information at the most appropriate time. Mixed-initiative approaches promise to dramatically enhance human-computer interaction by allowing agents to resemble human assistants. In this paper, we report a study on how the interaction metaphor can affect the user perception of agent capabilities and, in turn, the final success of agents.

Social group recommendation in the tourism domain

Recommender Systems learn users’ preferences and tastes in different domains to suggest potentially interesting items to users. Group Recommender Systems generate recommendations that intend to satisfy a group of users as a whole, instead of individual users. In this article, we present a social based approach for recommender systems in the tourism domain, which builds a group profile by analyzing not only users’ preferences, but also the social relationships between members of a group. This aspect is a hot research topic in the recommender systems area. In addition, to generate the individual and group recommendations our approach uses a hybrid technique that combines three well-known filtering techniques: collaborative, content-based and demographic filtering. In this way, the disadvantages of one technique are overcome by the others. Our approach was materialized in a recommender system named Hermes, which suggests tourist attractions to both individuals and groups of users. We have obtained promising results when comparing our approach with classic approaches to generate recommendations to individual users and groups. These results suggest that considering the type of users’ relationship to provide recommendations to groups leads to more accurate recommendations in the tourism domain. These findings can be helpful for recommender systems developers and for researchers in this area.

Building respectful interface agents

To provide personalized assistance to users, interface agents have to learn not only a users preferences and interests with respect to a software application, but also when and how the user prefers to be assisted. Interface agents have to detect the users intention to determine when to assist the user, and the users interaction and interruption preferences to provide the right type of assistance without hindering the users work. In this work we describe a user profiling approach that considers these issues within a user profile and a decision making approach that enables the agent to choose the best type of assistance for a given user in a given situation. We also describe the results obtained when evaluating our proposal in the tourism domain, and we compare these results with some previous ones in the calendar management domain.

Modeling sequences of user actions for statistical goal recognition

User goals are of major importance for an interface agent because they serve as a context to define what the user’s focus of attention is at a given moment. The user’s goals should be detected as soon as possible, after observing few user actions, in order to provide the user with timely assistance. In this article, we describe an approach for modeling and recognizing user goals from observed sequences of user actions by using Variable Order Markov models combined with an exponential moving average (EMA) on the prediction probabilities. The validity of our approach has been tested using data collected from real users in the Unix domain. The results obtained show that an interface agent can achieve near 90% average accuracy and over 58% online accuracy in predicting the most probable user goal after each observed action, in a time linear to the number of goals being modeled. We also found that the use of an EMA allows a faster convergence in the actual user goal.

NLP-based faceted search: Experience in the development of a science and technology search engine

An appropriate promotion, distribution and dissemination of scientific, artistic and technology developments can foster the collaboration between a countrys productive and academic sectors. The purpose of this paper is to present a novel search engine aiming at helping people to access science and technology advances, researchers and institutions working in specific areas of research. Our search engine first collects information disseminated on the Web in academic institution sites and in researchers personal homepages. Then, after intensive text processing, it summarizes the information in an enriched and user-friendly presentation oriented to non-expert users. Stable performance and an acceptable level of effectiveness for automatic named entities recognition indicate the potential of our approach for bridging the gap between the heterogeneous and unstructured information available on the Web about the research and development advances in a country and the innovation required by the productive sectors.

Personalized detection of user intentions

Interface agents are strategic software components for improving the quality of services to users. In order to be accepted by users, interface agents need to make useful suggestions always in the context of the users intention. The users intention should be detected as soon as possible so that the agent can define a way to collaborate with the user. Plan recognition can be applied to identify the users goal based on his or her actions in the environment. However, classical approaches to plan recognition fail in two main aspects that make them unsuitable for being used by interface agents: the lack of personalization and the lack of consideration of the transition between different goals pursued by the user. We propose an approach to capture intentions taking into account the variables involved in the application domain that represent the user preferences. Experimental evaluations show us that we have found a way for early detection of intentions.

Enhancing the experience of users regarding the email classification task using labels

Email is an indispensable tool for communication and users might have to deal with large volumes of information which they cannot always operate efficiently. For these users, the organization of emails is a tedious task. The use of automatic filters is not always possible or effective, because of difficulties regarding how to create a specific rule or because their use is impractical in some situations. In this article, we present an approach to enhance a webmail client with an interface agent that helps the user to label incoming email based on the knowledge of the users preferences. We not only considered the label that can be applied to different emails but also how to better interact with the user to provide him/her with assistance in the labeling procedure. We performed a set of experiments using Googles webmail system, Gmail, obtaining a good rate of acceptance of the agent interactions.