The Laboratory of Artificial Intelligence and Computer Science (LIACC) at the University of Porto was organized since its creation in 1988 (5 PhDs) as a federation of 3 groups that initially shared a common interest in Logic Programming and that were formed by researchers from 3 Faculties: Science, Economics and Engineering. This model proved to be quite adequate until recently (39 PhDs) when it became clear that the groups were growing in directions that instead of being based on research strategies defined by LIACC mainly depended on the teaching needs of the Faculties. Internal discussions on this issue started on May 2006. Our Advisory Board during its visit on October was asked to give its opinion (see section 2 of their report). The outcome of this process was 1) the spinning off of the AI and Data Analysis Group to form a new centre, and about one third of the Computer Science group to go to existing units; 2) a reorganization of LIACC (21 PhDs) as a single unit with 4 thematic research groups, working in collaboration on a few long-term research goals.

This reorganization created the opportunity for a definition of new long-term objectives, selected by their relation to hard real-world problems posed by the use of computers, and for increasing the cohesion of LIACC as a whole by linking these objectives to the short-term objectives of the research groups.

The management of LIACC is done according to the FCT Multi-year Funding Program rules, through a Coordinating Board, a Scientific Council and an external Advisory Board. The Coordinating Board is composed by three senior researchers, one of which is the LIACC Coordinator, and three researchers usually elected among the heads of the research groups. The Scientific Council, responsible for the approval of annual reports and plans, is composed by all the researchers holding a PhD. LIACC activities are periodically reviewed by an Advisory Board whose members are invited among internationally recognized outstanding experts.

Until December 2009 the following researchers will be in charge:

- LIACC Coordinator: Eugénio Oliveira

- LIACC Coordinating Board: Eugénio Oliveira, Luís Damas, Miguel Filgueiras, Armando Matos, Ana Paula Tomás, Luís Paulo Reis

- Advisory Board: Claude Kirchner (INRIA, France), Carles Sierra (IIA, CSIC, Barcelona), and new member to be confirmed

LIACC aims at helping to solve the following general problems:

- Software reliability

Reliable software is still the main challenge in computer science; software engineering approaches failed in general; formal approaches based on logic specifications and theorem proving have trouble handling real programs with a loose specification; unit tests improve the reliability of parts of the program but fail to guarantee reliability of the whole; design by contract approaches (e.g. ruby, spec-#) combined with type inference can improve reliability but do not guarantee correctness; may be a combination of different approaches should be used.

- Distributed, Dynamic and Decentralized problems

There are classes of problems that are inherently of a distributed, dynamic and decentralized nature as those which occur when heterogeneous partners need to collaborate over a computer network. Such a class comprises Electronic Institution problems related to services like negotiation mediation, ontology mapping, contract drafting, contract monitoring and enforcement, and evaluation of reputation indices for Virtual Enterprises life-cycle automatic assistance. These problems are better addressed through Autonomous Agents and Multi-Agent Systems (MAS). However improvements are required both on the theoretical and practical aspects, in order to achieve better interoperability, decision making, agents- and team- coordination. Approaches to this include the use of Adaptation, joint Learning capabilities, Emotion-like architectures for cognitive agents, agent-based simulation platforms. Direct application areas are Electronic Institutions problems and Intelligent Text Mining.

- Security

Security of software and of communication is currently a major concern. Public key cryptography techniques are based on the computational intractability of the associated code breaking problem. However this is normally judged on worst case complexity and better measures of security are needed. One possible approach is the use of time bounded versions of Kolmogorov complexity. Another problem arises from the need to execute foreign code that may open the door to attacks. One solution is to force foreign code to include a proof that it will not break security. This however requires support for the development of proofs of program properties.

- Information processing reliability

Representation of information is a major problem of computer science. Coherence and reliability of information is still almost completely left to programmers; e.g. information migration between representations is done without enforcing any kind of reversibility and can lead to loss of information; standards like XML ignore this kind of problem. There is a strong need for models (e.g. like reversible languages) and tools to achieve reliability in information processing.

- Results

A new technique based on Algorithmic Information Theory for the automatic classification of the fetal heart rate records important in detecting abnormal situations where the life of the fetuses may be at risk; the technique correctly clustered together and detected all the abnormal or suspicious tracings in an experiment with 31 fetuses

Prototype of Agent-based software platform for Virtual Enterprises life cycle, with support for Adaptive Negotiation protocol, Ontology-based services and preliminary Electronic Contracting. Software broker for the insurance domain using the same negotiation protocol

Significant contribution to the development of two sub-areas of Machine Learning: Meta-learning and Learning from data streams; this led to the edition of 4 special issues of journals

Design of new algorithms for matching problems with priorities, preferences and tenants. Research was triggered by flaws detected in the Portuguese hiring law for state-school teachers. Results sent to the Government were taken in a new law.

Agent-based common framework suitable for controlling teams of cooperative robots for Robosoccer (with IEETA); competitions: Simulation League 3D (World Champion 2006; European Champion 2006); Simulation League 2D ( 2nd, European contest 2006); Coach Competition (2nd, World contest 2003, 2nd, 2004); Rescue Simulation League (European Champion 2006)

Contributions to the area of Data Mining using ILP: speed up the execution of queries by transformations based on program analysis, tabling techniques, deductive DB framework for efficient and scalable induction of logic programs, APRIL an ILP system

A complete characterization of the maximum transmission rate at which the eavesdropper is unable to decode any information when 2 partners communicate over a quasi-static fading channel and an eavesdropper observes transmissions through another quasi-static fading channel (with Cambridge Univ., Georgia Tech)

A complete formal definition of the Formula-Tree Proof Method for investigating type inhabitation together with a proof of correctness. This method is a valuable tool for research in the area, and fostered a large amount of new results

An architecture for Emotion-like based Agents. Simulation of a practical application in firefighting and a new logic formalism for specifying simultaneously guided E_BDI Agents by Beliefs, Desires, Intentions plus Emotion-like mechanisms

Yap kept as a state-of-the-art Prolog implementation with support for constraints, tabling, parallelism, external data-bases, and spatial data-bases

A language whose programs characterizes a fundamental and powerful set of reversible transformations on integer tuples. This is similar to the characterization of primitive recursive functions by a simple programming language

Identification of circumstances under which Multi-Agent Systems joint learning through advice is effective

A unique string representation for complete ICD finite automata, shown to be adequate for exact/uniform random generation, providing an alternative way of counting and an optimal coding. Exact/approximate values for the number of regular languages and density of minimal automata were obtained with an impact in average-case complexity studies

The work on Computational Expressivity of Linear Calculi by Sandra Alves was awarded a Gulbenkian Research Stimulus Prize 2005

MAS for a real application on Operations control recovery for crew and plane scheduling

- Other strengths

links to important groups abroad

list of publications

list of prototypes

- Research Units in other disciplines with which there were joint projects:

CEMAS, Porto [Environmental systems, Agent-based simulation]

IBMC, Porto [Bioinformatics, Data Mining]

SINTEF, Oslo [Text information extraction and retrieval]

ISR, Porto [Hardware, Automated guided vehicles]

Universidade de Coimbra, Faculdade de Psicologia e Ciências da Educação [Handling of structured documents]

Universidade do Porto, Faculdade de Ciências, Theoretical Chemistry Group [Protein segmentation, ILP]

Universidade do Porto, Faculdade de Letras [Handling of structured documents]

Universidade do Porto, Faculdade de Medicina [Medical Informatics]

Universidade do Porto, Faculdade de Psicologia e Ciências da Educação [Handling of structured documents]

- collaboration in postgraduate actions

responsibility for Doctoral PhD Programme on Informatics Engineering, University of Porto

set up of the Doctoral PhD Programme on Computer Science, Universities of Aveiro, Minho and Porto

Master courses at University of Porto:

* Informatics

* Informatics Engineering

* Artificial Intelligence and Intelligent Systems

- collaboration in other formation actions:

European Join Robolution BEST course

supervision of EU-Socrates scholarship holders

- links with public-sector and industrial partners

INE

Intevation Gmbh

Parque Natural da Ria Formosa

PortalExecutivo.com

Sensoria

SAS Portugal

STCP

TAP - Transportes Aéreos Portugueses

WhiteBook Consulting

- software with wide distribution

Apoo: an Environment for a First Course in Assembly Language Programming, 1998-2004. Linux Debian and Ubuntu distributions

GPSMan: a graphical manager of GPS data, 1998-2006, Linux Debian, Ubuntu and FreeBSD distributions; part of the Free-GIS CD

Mooshak: a system for managing programming contests on the Web, 2002-2005. Used in ACM programming competitions

Yap: state-of-the-art Prolog implementation with support for constraints, tabling, parallelism, external data-bases, and spatial data-bases. Available from Sourceforge

- systems for disabled people

starting the development of a wheelchair with intelligent control

development of a translator from mathematical expressions in LaTeX into Braille, Portuguese variant

- science dissemination

collaboration with Atractor, an association for the diffusion of Mathematics, in the development of interactive modules

several seminars on AI and Intelligent Robotics given at high-schools

What must be expected from LIACC is to continue to be one of the Portuguese topmost research units in the areas of Theoretical Computer Science, Multi-agent Systems, and Programming Languages Implementation, with a good level of international recognition. The impact of outreach activities must be increased, if not by altruistic reasons, as a means of gathering case studies and achieving better external visibility.

With these goals in mind a SWOT (Strengths, Weaknesses, Opportunities and Threats) was performed that can be summed up as follows:

- Strengths: strong research record of the current staff, including publications; broad coverage of subjects by staff members; technological competence on relevant application areas; low financial running costs; team cohesion has recently increased

- Weaknesses: staff members tend to address too much research topics; poor level of recognition by non-specialist scientific community and by the general public; no own premises and dispersion of groups in different campuses

- Opportunities: recent reorganization allowed for a redesign of goals and strategies, as well as for the existence of quality-ensuring mechanisms; new PhD programme in collaboration with other universities; funding for hiring post-docs; funding for research projects; links with both research units in other disciplines, and public and industry partners

- Threats: recruiting problems at post-graduate level, with few acceptable candidates; funding problems due to delays on transfers making impossible the existence of spending plans

The following actions should be taken in view of this analysis:

1. to seek partnerships with foreign internationally recognized research units for exchange of post-doctoral researchers, and launching of joint projects

2. collaboration in post-graduation programmes with foreign outstanding universities

3. to establish further links with important public-sector and industrial partners increasing the number of results on real-world problems, while preventing from becoming a service provider of standard technologies

4. to foster promising research areas after a careful evaluation of intermediate results; possible candidates may include Medical Informatics, Intelligent Traffic Systems and Information Extraction from Corpora

5. to reinforce the existing teams and counter the lack of staff at the PhD level by recruiting young post-docs

6. to make efforts in finding commonalities and possibly merge with adequate high quality research units working in close topics

7. to increase LIACC visibility by producing applications with a large

potential number of users both in the academic and non-academic communities,

which should be showcases of advanced programming systems

CCTC - Centro de Ciências e Tecnologias de Computação, Theory and Formal Methods Group, Prof. José Nuno Oliveira, Prof. Jorge Pinto

CEOC - Centro de Optimização e Controlo, Prof. António Bajuelos

CEREM - Universidade Fernando Pessoa, Prof. Nuno Ribeiro

CISTER - Centro de Investigação em Sistemas Confiáveis e de Tempo Real, Prof. Luís Pinho

CMUP - Centro de Matemática da Universidade do Porto, Prof. António Machiavello, Prof. Manuel Delgado

IEETA - Instituto de Engenharia Electrónica e Telemática de Aveiro, Prof. Nuno Lau

ISR - Instituto de Sistemas e Robótica, Porto, Prof. Lobo Pereira

INESC Porto - Prof. Augusto de Sousa

IPB- Escola Superior de Tecnologia e Gestão, Instituto Politécnico de Bragança, Prof. Paulo Leitão.

IT - Instituto de Telecomunicações, Prof. Amílcar Sernadas (Lisboa), Prof. João Barros (Porto)

TAP - Transportes Aéreos Portugueses, Lisboa

UNINOVA - Prof. Steiger Garção

Universidade de Coimbra, Faculdade de Psicologia e Ciências da Educação, Prof. Armando Luís Dinis, Mónica de Oliveira

Universidade do Porto, Faculdade de Letras, Prof. Manuel Loff

Universidade do Porto, Faculdade de Medicina, Prof. Altamiro Costa Pereira

Universidade do Porto, Faculdade de Psicologia e Ciências da Educação, Dra. Teresa Medina, Dra. Cristina Nogueira