Aitor Almeida
Universidad de Deusto, DeustoTech, Department Member
Managing data within a research unit is not a trivial task due to the high number of entities to deal with: projects, researchers, publications, attended events, etc. When all these data are exposed on a public website, the need to have... more
Managing data within a research unit is not a trivial task due to the high number of entities to deal with: projects, researchers, publications, attended events, etc. When all these data are exposed on a public website, the need to have it updated is fundamental to avoid getting an incorrect impression of the group's performance. As research centres websites are usually quite static, external documents are generated by managers, resulting in data redundancy and out-of-date records. In this paper, we show our efforts to manage all these data using Labman, a web framework that deals with all the data, links entities and publishes them as Linked Open Data, allowing to get insightful information about the group's productivity using visual analytics and interactive charts.
Research Interests:
This article describes our system presented at the workshop for sentiment analysis TASS 2015. Our system approaches the task 1 of the workshop, which consists on performing an automatic sentiment analysis to determine the global polarity... more
This article describes our system presented at the workshop for sentiment analysis TASS 2015. Our system approaches the task 1 of the workshop, which consists on performing an automatic sentiment analysis to determine the global polarity of a set of tweets in Spanish. To do this, our system is based on a model supervised Linear Support Vector Machines combined with some polarity lexicons. The influence of the different linguistic features and the different sizes of n-grams in improving algorithm performance. Also the results obtained, the various tests that have been conducted, and a discussion of the results are presented.
Research Interests:
Research Interests:
Knowledge-driven activity recognition is an emerging and promising research area which has already shown very interesting features and advantages. However, there are also some drawbacks, such as the usage of generic and static activity... more
Knowledge-driven activity recognition is an emerging and promising research area which has already shown very interesting features and advantages. However, there are also some drawbacks, such as the usage of generic and static activity models. This paper presents an approach to using data-driven techniques to evolve knowledge-driven activity models with a user's behavioral data. The approach includes a novel clustering process where initial incomplete models developed through knowledge engineering are used to detect action clusters which represent activities and aggregate new actions. Based on those action clusters, a learning process is then designed to learn and model varying ways of performing activities in order to acquire complete and specialized activity models. The approach has been tested with real users' inputs, noisy sensors and demanding activity sequences. Initial results have shown that complete and specialized activity models are properly learned with success rates of 100% at the expense of learning some false positive models.
Research Interests:
Evaluating human activity recognition systems usually implies following expensive and time-consuming methodologies, where experiments with humans are run with the consequent ethical and legal issues. We propose a novel evaluation... more
Evaluating human activity recognition systems usually implies following expensive and time-consuming methodologies, where experiments with humans are run with the consequent ethical and legal issues. We propose a novel evaluation methodology to overcome the enumerated problems, which is based on surveys for users and a synthetic dataset generator tool. Surveys allow capturing how different users perform activities of daily living, while the synthetic dataset generator is used to create properly labelled activity datasets modelled with the information extracted from surveys. Important aspects, such as sensor noise, varying time lapses and user erratic behaviour, can also be simulated using the tool. The proposed methodology is shown to have very important advantages that allow researchers to carry out their work more efficiently. To evaluate the approach, a synthetic dataset generated following the proposed methodology is compared to a real dataset computing the similarity between sensor occurrence frequencies. It is concluded that the similarity between both datasets is more than significant.
Research Interests:
When performing analytics on educational datasets, the best scenario is where the dataset was designed to be analyzed. However, this is often not the case and the data extraction becomes more complicated. This contribution is focused on... more
When performing analytics on educational datasets, the best scenario is where the dataset was designed to be analyzed. However, this is often not the case and the data extraction becomes more complicated. This contribution is focused on extracting social networks from a dataset which was not adapted for this type of extraction and where there was no relation among students: a set of remote laboratories where students individually test their experiments by submitting their data to a real remote device. By checking which files are shared among students and submitted individually by them, it is possible to know who is sharing how many files with who, automatically extracting what students are bigger sources. While it is impossible to extract the full real social network of these students, all the edges found are clearly part of it. These relations can indeed be used as a new input for performing the analytics on the dataset.
Research Interests:
ABSTRACT Purpose ‐ The purpose of this paper is to review the state-of-the-art in adaptive user interface systems by studying their historical development over the past 20 years. Moreover, this paper contributes with a specific model... more
ABSTRACT Purpose ‐ The purpose of this paper is to review the state-of-the-art in adaptive user interface systems by studying their historical development over the past 20 years. Moreover, this paper contributes with a specific model combining three main entities (users, context and devices) that have been demonstrated to be always represented in these environments. Novel concepts that should be taken into account in these systems are also presented. Design/methodology/approach ‐ The authors first provide a review and a comparison of current user interface adaptive systems. Next, the authors detail the most significant models and the set of techniques used to, finally, propose a novel model based on the studied literature. Findings ‐ Literature solutions for adaptive user interface systems tend to be very domain dependant. This situation restricts the possibility of sharing and exporting the information between such systems. Furthermore, the studied approaches barely highlight the dynamism of these models. Originality/value ‐ The paper is a review of adaptive user interface systems and models. Although there are several reviews in this area, there is a lack of research for modelling users, context and devices simultaneously in this domain. The paper also presents several significant concepts that should be taken into account to bring an adaptive and dynamic perspective to the studied models.
Research Interests:
ABSTRACT Knowledge-driven activity recognition is an emerging and promising research area which has already shown very interesting features and advantages. However, there are also some drawbacks, such as the usage of generic and static... more
ABSTRACT Knowledge-driven activity recognition is an emerging and promising research area which has already shown very interesting features and advantages. However, there are also some drawbacks, such as the usage of generic and static activity models. This paper presents an approach to using data-driven techniques to evolve knowledge-driven activity models with a user’s behavioral data. The approach includes a novel clustering process where initial incomplete models developed through knowledge engineering are used to detect action clusters which represent activities and aggregate new actions. Based on those action clusters, a learning process is then designed to learn and model varying ways of performing activities in order to acquire complete and specialized activity models. The approach has been tested with real users’ inputs, noisy sensors and demanding activity sequences. Initial results have shown that complete and specialized activity models are properly learned with success rates of 100% at the expense of learning some false positive models.
Research Interests:
ABSTRACT Adaptive user interfaces involves the design of dynamic interfaces, the main purpose of which is to present an adapted alternative to the user to ease the interaction. User's preferences, context situation, and... more
ABSTRACT Adaptive user interfaces involves the design of dynamic interfaces, the main purpose of which is to present an adapted alternative to the user to ease the interaction. User's preferences, context situation, and device's capabilities help these systems to adapt the interface to make the interaction more adequate to the current situation. Being aware of different characteristics of these entities is vital for reaching the main goals of these systems efficiently. To collect knowledge from these entities, it is necessary to design several formal models to help to organize and give meaning to the gathered data. This article analyzes several literature solutions for modeling users, context, and devices considering different approaches. The article identifies their advantages and drawbacks to finally propose a new ontology model that addresses the identified limitations.
Research Interests:
The participation of users within AAL environments is increasing thanks to the capabilities of the current wearable devices. Furthermore, the significance of considering user's preferences, context conditions and... more
The participation of users within AAL environments is increasing thanks to the capabilities of the current wearable devices. Furthermore, the significance of considering user's preferences, context conditions and device's capabilities help smart environments to personalize services and resources for them. Being aware of different characteristics of the entities participating in these situations is vital for reaching the main goals of the corresponding systems efficiently. To collect different information from these entities, it is necessary to design several formal models which help designers to organize and give some meaning to the gathered data. In this paper, we analyze several literature solutions for modeling users, context and devices considering different approaches in the Ambient Assisted Living domain. Besides, we remark different ongoing standardization works in this area. We also discuss the used techniques, modeled characteristics and the advantages and drawbacks of each approach to finally draw several conclusions about the reviewed works.
Research Interests:
Abstract While creating a framework for adaptive mobile interfaces for m-learning applications we found that in order to ease the use of our framework we needed to present the mobile device characteristics to non-expert users in a easy to... more
Abstract While creating a framework for adaptive mobile interfaces for m-learning applications we found that in order to ease the use of our framework we needed to present the mobile device characteristics to non-expert users in a easy to understand manner. Using fuzzy sets to represent the characteristics of mobile devices, non-expert developers such as teachers or instructional designers can actively participate in the development or adaptation of the educational tools. To be able to automatically generate the fuzzy membership ...
Research Interests:
Abstract. Ontologies are a useful and attractive tool for representing knowledge. In fact, if all the documents in the web were represented with ontologies, the job of search engines, automatic document processors, etc. would be much... more
Abstract. Ontologies are a useful and attractive tool for representing knowledge. In fact, if all the documents in the web were represented with ontologies, the job of search engines, automatic document processors, etc. would be much easier. However, ontologies are too complex to be used by the general public and, so far, are used only by specialized users. Nonetheless, a more informal type of classifying resources is becoming increasingly popular amongst the general public: social tagging or folksonomies. Many popular websites (del. ...
Research Interests:
Research Interests:
Research Interests:
Abstract: To be able to react adequately a smart environment must be aware of the context and its changes. Modeling the context allows applications to better understand it and to adapt to its changes. In order to do this an appropriate... more
Abstract: To be able to react adequately a smart environment must be aware of the context and its changes. Modeling the context allows applications to better understand it and to adapt to its changes. In order to do this an appropriate formal representation method is needed. Ontologies have proven themselves to be one of the best tools to do it. Semantic inference provides a powerful framework to reason over the context data. But there are some problems with this approach.
Recommender systems have increased their impact in the Internet due to the unmanageable amount of items that users can find in the Web. This way, many algorithms have emerged filtering those items which best fit into users' tastes.... more
Recommender systems have increased their impact in the Internet due to the unmanageable amount of items that users can find in the Web. This way, many algorithms have emerged filtering those items which best fit into users' tastes. Nevertheless, these systems suffer from the same shortcoming: the lack of new user data to recommend any item based on their tastes.
Intelligent environments offer information filled spaces. When trying to navigate among all the offered resources users can be overwhelmed. This problem is increased by the heterogeneous nature of resources in smart environments. Users... more
Intelligent environments offer information filled spaces. When trying to navigate among all the offered resources users can be overwhelmed. This problem is increased by the heterogeneous nature of resources in smart environments. Users must choose between a plethora of services, multimedia information, interaction modalities and devices. But at the same time the unique characteristics of smart spaces offers us more opportunities to filter these resources.
Abstract: The number of resources (services, data, multimedia content, etc) available in Smart Spaces can ver overwhelming. Finding the desired resource can be a tedious and difficult task. In order to solve this problem, Smart Spaces... more
Abstract: The number of resources (services, data, multimedia content, etc) available in Smart Spaces can ver overwhelming. Finding the desired resource can be a tedious and difficult task. In order to solve this problem, Smart Spaces contain much information that can be employed to filter these resources. Using the user context-data available in Smart Spaces can help refining and enhancing the recommendation process, providing more relevant results.
Abstract This work describes an OSGi-based middleware platform to enable more scalable, future-proof, cost-efficient and standard-following intelligent environments. It complements OSGi with two main features which make it even more... more
Abstract This work describes an OSGi-based middleware platform to enable more scalable, future-proof, cost-efficient and standard-following intelligent environments. It complements OSGi with two main features which make it even more suitable for intelligent environment management: a) dynamic discovery and monitoring of distributed semantic services and b) semantic context modelling and reasoning for intelligent service provision.
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Being able to recognise human activities by means of sensor and computational devices can be a key competence in order to achieve human centred technologies. For that purpose, it is mandatory to build computational models of the... more
Being able to recognise human activities by means of sensor and computational devices can be a key competence in order to achieve human centred technologies. For that purpose, it is mandatory to build computational models of the activities which have to be recognised. There are two major approaches for activity modelling: the data-driven and the knowledge-driven approaches. Both of them have advantages and drawbacks. The objective of this work is to combine both modelling approaches with the aim of building dynamic and personalised activity models, using generic knowledge-based models. This would allow implementing modelling processes which can adapt themselves to the evolution of specific people.
Being able to recognise human activities by means of sensor and computational devices can be a key competence in order to achieve human centred technologies. For that purpose, it is mandatory to build computational models of the... more
Being able to recognise human activities by means of sensor and computational devices can be a key competence in order to achieve human centred technologies. For that purpose, it is mandatory to build computational models of the activities which have to be recognised. There are two major approaches for activity modelling: the data-driven and the knowledge-driven approaches. Both of them have advantages and drawbacks. The objective of this work is to combine both modelling approaches with the aim of building dynamic and personalised activity models, using generic knowledge-based models. This would allow implementing modelling processes which can adapt themselves to the evolution of specific people.
Effective handling of research related data is an ambitious goal, as many data entities need to be suitably designed in order to model the distinctive features of different knowledge areas: publications, projects, people, events and so... more
Effective handling of research related data is an ambitious goal, as many data entities need to be suitably designed in order to model the distinctive features of different knowledge areas: publications, projects, people, events and so on. A well designed information architecture prevents errors due to data redundancy, outdated records or poor provenance, allowing both internal staff and third parties reuse the information produced by the research centre. Moreover, making the data available through a public, Internet accessible portal increases the visibility of the institution, fostering new collaborations with external centres. However, the lack of a common structure when describing research data might prevent non-expert users from using these data. Thus we present labman, a web-based information research system that connects all the actors in the research landscape in an interoperable manner, using metadata and semantic descriptions to enrich the stored data. Labman presents different visualizations to allow data exploration and discovery in an interactive fashion, relying on humans’ visual capacity rather than an extensive knowledge on the research field itself. Thanks to the visual representations, visitors can quickly understand the performance of experts, project outcomes, publication trajectory and so forth.