PROGRAMMING LANGUAGES

Author: PINTO ALARCÓN MÓNICA.
Year: 2003.
University: MÁLAGA [www.uma.es].
Place of defense: ETSI INFORMÁTICA.
Place of preparation: UNIVERSIDAD DE MÁLAGA.

Summary: Continued technological advances, both of the systems hardware and communications technologies have led to an increase in the development of heterogeneous distributed systems and open, it has become necessary to the adoption of new software technologies. Two of these technologies are the Software Development Based on Components (DSBC) and the Software Development Oriented aspects (DSOA), which have emerged with the aim of improving the modularity and evolucióon of systems software by assembling and quantities indpendientes reusable. The DSBC and DSOA are complementary technologies, based both in the implementation of the principle of separation of advanced concepts from the perspective diferntes, and consequently serí very helpful to have models that combine these technologies to get together all of their benefits. The main objective of this thesis is to integrate the principles of DSBC and DSOA by defining d eun model, a language and platform components and aspects. The model CAM (Component-Aspect Model) aims to solucinar limitations of the platforms of components and systems for current issues, defining the components and entities as aspects of the first order, with a mechanism composition of components and appearance are intrusive. In addition, the model allows CAM defined as both aspects common services delas distributed applications like any other code replicated or entangled, defining a model of homogeneous composition. Once the model CAM, another very important feature of CAM / DAOP is that the information on the architecture of the application is not distributed among the components and aspects of the implementation, as usual. Instead, this information is definie explicitly using a description language architecture called DAOP-ADL. Like similar LDAs, this language is used pair analyze and verify the correctness of a certain design, and thinking about the possible structure of the system and evaluate various design options, even before its implementation. DAOP-ADL has been described using XML schemas, which allow the information provided with the language can be interpreted and used directly by the platform DAOP. The platform DAOP (Dynamic Aspect-Oriented Platform) is the infrastructure on which will be implemented components and aspects CAM. The most important service offered by the platform DAOP is to compose the components and aspects, or tejerlos if we use the terminology of aspects of dynamic execution times. To establish connections between components and aspects, the platform DAOP check information ddescrita anterioremtne with DAOP-ADL, which will be stored in their internal structures. Thus, it is possible to adjust the number and type of aspects tissue with a particular component, even at runtime, improving the flexibility and adaptability of the final application.

Author: RODRÍGUEZ LAGUNA ISMAEL.
Year: 2003.
University: COMPLUTENSE DE MADRID [www.ucm.es].
Place of defense: FACULTAD DE INFORMÁTICA.
Place of preparation: FACULTAD DE INFORMÁTICA.

Summary: The problem of the distribution of computational resources between competing entities in a distributed system is complex due to several reasons. On the one hand, neceistamos satisfy, as far as possible, different goals that can be mixed and inconsistent. Moreover, we need to obtain appropriate distributions using a small number of messages not overload communications system. Science provides the economic concepts needed to treat environments where a number of players competing for the use of certain scarce resources that could have alternative uses, so its application to this problem is adequate. On the other hand, formal methods take years to abstraction providing adequate mechanisms to analyze the properties of computer systems. Specifically, the language specification of concurrent systems allow for formal definition of distributed systems and subsequent studies of its properties in a formal way. In this thesis is built on formal language PAMR (Process Algebra for the Management of Resources), which allows specifying concurrent systems where the availability of resources among different processes play an important role. The definition of the relevant characteristics of the systems is done using concepts of Microeconomics, which allows use of this technique in the analysis of such systems. Moreover, the language is extended to include an incremental factors have a strong correlation with the amount of resources available. Such is the case of time, whose influence on the behavior of the systems is discussed either version deterministic (ie you can define the time it takes each task depending on the resources available) and its probabilistic version (ie, aware of the likelihood that a task afternoon, a time determined by the resources available). Moreover, the handling of the economic concepts of our language will be appropriate for use in a different field of Information Technology, Electronic Commerce. While in that area available resources should be distributed among end users human rather than between computational entities, the techniques necessary for the formal description of such environments in economic terms will be similar to ls that usáraos in the description of distributed systems. A new extension of our language allows the narrative and formal analysis of such systems. Both in the case of this extension, as in the case of those who seek the introduction of time, our formal language can be inferred in a systematic manner to those behaviors that are compliant with the specification of a given system, and distinguish them from those who do not conduct they are. For those present metodologíasde testing extensions that provide systematic techniques to check whether a given implementation meets the behavior defined by a specification. The explicit description of the dependence on resources will force the construction of methodologies that take into account the influence on the behavior of systems.

Author: SANCHEZ ALONSO M. SOLEDAD.
Year: 2003.
University: EXTREMADURA [www.unex.es].
Place of defense: ESCUELA POLITÉCNICA.
Place of preparation: ESCUELA POLITECNICA.

Summary: This thesis presents (OFRE Coordinativa Formal Requirenets Encinooment), an environment for the development of coordinated environments based components where the functional aspects and coordination are regulated separately from the early stages. The environment that is proposed is based on -- the formal language sends as a basis for especificacción and development on the part of the tools of the environment, and use of language exogenous coordination Coordenate the Poles.

Author: MARTÍNEZ CRUZ JESÚS.
Year: 2004.
University: MÁLAGA [www.uma.es].
Place of defense: ETSI INFORMATICA.
Place of preparation: E.T.S.I. INFORMATICA.

Summary: In this paper, thesis deals with the use of XML and MDA technologies for the integration of analysis tools portocolos.El core of the proposal is an XML to specify protocols called PiXL (Protocol With an XML Interchange Language), which has allowed the implementation of applications with relative ease verification through automatic checking abstract models, reusing completely model checking tools and adding modules for data abstraction and eventos.El language and methodologies proposals also apply to the type extension of análisis.En particular, the concepts introduced also apply to the possibility of integrating performance analysis protocols in the context of the Active Network.

Author: AMOR PINILLA M. DE LAS MERCEDES.
Year: 2004.
University: MÁLAGA [www.uma.es].
Place of defense: E.T.S.I. INFORMÁTICA.
Place of preparation: E.T.S.I. INFORMÁTICA.

Summary: The main contribution of the model agent Malaga is simplifying the process of developing software agents, giving solution to the constraints affecting the work of APLs and frameworks Object Oriented existing building systems Multi-Agente (SMA). These limitations stem from the use of OOP design inflexible and the presence of problems such as code entangled and replication of code, in various components of the internal architecture of the agent. These problems affect reuse and adaptability of the internal structure of the players, complicating their construction and hamper their subsequent evolution. The main contribution of this thesis has been to combine the principles of Software Development Based on Components and Software Development Oriented aspects in the definition of Malacca, a model and an agent software architecture aimed at improving the development and evolution of GHS resolved before Reviewed constraints. Within the software architecture of an agent and applying the DSBCla functionality dependent on a domain of application is provided by COTS components. This promotes reuse and prevents the development of software agents from scratch reducing the cost, time and effort devoted to implementation process. The implementation of DSOA allows us to get a better descomposicíon functional dela internal architecture of the agent. To this end we identified and separated into separate entities within the architecture any property belonging to the reporting agent, which is modeled as an aspect. This identification led us to separate and model as the coordination aspects according to a protocol interaction, encryption of messages ACL in different performances and the distribution of messages through the services provided by a platform agents. Generally agents are arising more extensible, reusable and adaptable because its internal architecture allows you to add, delete, or replace components and aspects without the other components and aspects of the agent is affected. This feature provides significant benefits, as the training of new protocols at run time and access, even simultaneously, in different platforms to communicate IFAP agents. These and other interesting contributions arising from the characteristics of the model are illustrated along the memory. Along with the model are built two languages specific domain whose aim is to simplify and facilitate the development and evolution of the agents built on the architecture Malacca. For a description dela coordinating officers has been defined ProtDL, a language for specifying protocols interaction between players based on XML. Thus, the interaction protocols supported by an actor is represented explicitly, facilitating their integration and modification. In addition to the description of new players has been defined MaDL, a language for describing architectures agents based components and aspects. Thus, the developer should only provide a description of the components to be assembled at the agent, an explicit description of the interaction protocols supported ProtDL, and information about the configuration and deployment of other aspects of the agent in an XML document . Finally, another contribution of this dissertation is the integration of Malacca model in the development process-oriented methodology proposed by agents through the application and MDA. The model integrates Malacca in a development process guided by models characterized by promoting the use of any methodology agents. The application of MDA reduces the "jump" between design and implementation and to solve the variety of methodologies, facilitand 8 or inclus 1c4 or automating last step of development.

Author: PLAZA GARCIA INMACULADA.
Year: 2004.
University: ZARAGOZA [www.unizar.es].
Place of defense: E.U. DE INGE. TEC. INDUSTRIAL.
Place of preparation: CENTRO POLITECNICO SUPERIOR.

Summary: This thesis aims at the development of a generic PLC programmable (PLC) - based real-time operating systems (RTOS) under quality criteria. A previous study of the commercial products lets look at some trends, such as the quest for quality and standardization. In particular, manufacturers mentioned the IEC 61131-3:2003 describing programming languages such devices. However, a study by the state of the art allows us to identify gaps in this area. In particular, lack a technical description of the rule-clarifying ambiguous points or imprecisos- and a free tool for compilation. On the other hand, work according to quality criteria requires not only focus on the product, but also address aspects of the development process itself. Attempting to respond to these needs has been raised a division of the theory in two large blocks: Block-Quality: it examines the desirability of introducing a system of quality management in the group of university R & D and innovation in the who is going to develop the thesis work. Given the positive response, explores different regulations, eventually designing a quality management system suitable for this type of organization. Its presence in the group has also identified a new line of work that we have called "Quality in Education", to implement the concepts learned about quality in our teaching. - Block Model Autómata: in which analyzes the IEC above, indicating ambiguous points and suggestions for improvement. Once clarified theoretical concepts proposes an implementation of the Model Software IEC 61131-3 with RTOS. It also presents the implementation of a language translator List Briefs (one of those described in the rule) BC, which was developed under the guidelines of the management system of quality defined in the previous block. The model has been implemented in developed PC platform, using as operating system RTLinux. As we highlight features that broadens the potential user regarding the PLC business. Likewise, the execution times are comparable to those of other PLC-based PC platform that can be found in the market. As additional advantages, we will note that the software developed favors a structured programming, allowing a strong data typing and control of the execution, gives support for structured data and also favors an independence of the seller. These qualities have been remarcadas in literature as features that should be considered for PLC software quality. Finally, note that the tool is developed independent (no need to integrate into other applications software development), allows the entry in a file editable using any text editor, does not use any specific instruction of a specific manufacturer and is completely based FOSS

Author: GARRIDO MÁRQUEZ DANIEL.
Year: 2005.
University: MÁLAGA [www.uma.es].
Place of defense: E.T.S.I. INFORMÁTICA.
Place of preparation: E.T.S.I. INFORMÁTICA.

Summary: The work of the thesis has focused on the integration of three components: software components, middleware and analysis planificabilidad. These three elements combined allow for the creation of distributed systems in real time, ie distributed systems predictable and analysable. It has created a model of software components suitable for real-time systems. This model overcomes the shortcomings of the standard models of components. The two main advantages of the model are other approaches, obtaining an abstract model SDL component and the use of annotations for the calculation of the worst times of execution. The combination of these two elements is used to obtain independence platform. The second part of the work of the thesis has focused on the distribution of the standard components on a platform of implementing predictable. Among the many possibilities, it was decided to use RT-CORBA by the many advantages that this middleware provides for the realization of real-time distributed applications. The third element has been the creation of a methodology for the analysis of planificabilidad of applications based on this model components and on this platform communications predictable. Finally, one of the main objectives is the implementation in real systems. We have used the proposals made in prototype simulators nuclear power plants, checking the suitability of such proposals.

Author: GUZMÁN DE LOS RISCOS EDUARDO.
Year: 2005.
University: MÁLAGA [www.uma.es].
Place of defense: E.T.S.I INFORMÁTICA.
Place of preparation: E.T.S.I. INFORMÁTICA.

Summary: The Guardians Intelligent Systems (ITS), are tools for teaching that apply artificial intelligence techniques to guide the student during the process of investigation. They use a model of the student that represents what it knows and what it does not know about the domain being instructed. This model is updated during the process of investigation and it is essential to determine at any given time what should be done. The structure that stores the state of knowledge of the learner is the proper model pupil, while the process of reasoning which updates this model is called the student's diagnosis. The importance of an accurate diagnosis and rigorous is vital to the smooth functioning of an STI. A test is an assessment tool designed to infer a measure of the capabilities of subjects, through their responses to a set of questions (or items). Within the diagnosis in STI, the tests have become a popular thanks to the benefits of its use. These highlights the fact that they are applicable to a greater or lesser degree in virtually all domains. However, the way in which they built the way it is applied often, usually incorrect, or at least little harsh. Like any other measuring instrument, it must require certain characteristics, through which it can verify their suitability. The Tests Adaptive Informatizados (TAI), are behind a solid theoretical foundation that ensures the validity, reliability and objectivity of the assessment being done. Through a TAI the result is the estimate of the level of knowledge of the student. They require fewer test questions with respect to the conventional tests. This is because each question is selected according dela estimate that so far has done the test of knowledge of the student. The TAI is based on applying an algorithm for evaluation, which is based primarily on a theory piscométrica called Item Response Theory (TRI). According to this, that a considered response to a given question, is linked to the level of knowledge that he possesses. This relationship is quantified through one or more functions called density characteristics curves. Despite its suitability for the diagnosis of knowledge, TAI presented a set of drawbacks. The first is that only can offer as a result, in each test, a single assessment of the student's knowledge. Moreover, in those tests in which they are involved questions about various concepts, the selection criteria often resort to heuristic, which casts doubt on the thoroughness of the results. Another problem of the TAI is that they are very expensive, in terms of development time. Before a TAI is operational, it is necessary to calibrate cornering characteristics of each question. In general, the procedures that exist for this purpose are very expensive, since they require to have examined a sample population and size considerably. In the field of STI, there are various proposals under which makes use of the TAI but, in general, do not use properly the features and benefits they can offer diagnostic knowledge. Models diagnostic cognitive pursue a more ambitious target that TAI, as they try to provide a set of inferences most sophisticated numerical estimate that the mere knowledge of the learner is a unique concept that provides a TAI. The solution proposed in this thesis is based on the definition of a standard response based on the TRI and other diagnostic cognitive, whose evaluation tool are TAI. The proposal includes several features that enable solve the problems of TAI with regard to diagnosis in the STI. Also, the model r 8 espuesta 3f0 defined provides a theoretical foundation of the diagnosis. Allows combine in a single test different types of questions, and includes an algorithm for the calibration of them, saving the initial requirements that this kind of algorithms needed to be applied. Both the model answer as the diagnostic has been implemented in a system Web SIETTE, which is capable of operating as an independent assessment tool, or as a diagnostic module within the architecture of an STI, thanks to a set the protocols that have been implemented for this purpose.

Author: BERJÓN GALLINAS ROBERTO.
Year: 2005.
University: DEUSTO [www.deusto.es].
Place of defense: FACULTAD DE INGENIERIA.
Place of preparation: FACULTAD DE INGENIERIA.

Summary: Due to the need for companies not only to exchange information among themselves, but also to make it accessible to as many customers as possible, for example through the Web or mobile terminals, it is imperative that encrypt information using a set of patterns. On the one hand, codification must be compatible with all information systems to ensure their readability. It must also be verifiable to verify that its structure is correct. Moreover, it should not be a loss, that is the meaning of the information. Lastly, it must be easily convertible, in order to convert it to other formats. All these requirements have made XML a standard for encoding the information, because it meets all of these requirements. Therefore, companies increasingly require that your information is represented in this format. This is why we have recently presented new models of databases complementing the XML files and traditional relational databases, and allow the storage of information in XML bases native XML and relational databases enabled for XML. Today, companies are faced with their information distributed among various heterogeneous sources, which of course use different languages to access and are in different formats. All this makes it very difficult not only to find the information, but its acquisition in XML format. There are numerous works and studies that propose solutions to this problem: the search and retrieval of information in XML format. However, these efforts are concentrating solely on the processing of XML information residing in relational databases. As previously noted above, this is not enough. Requires integrate data from various sources, taking into account that can be of different types. Thus, the objective of this thesis is precisely this: to integrate information from heterogeneous data sources via XML. This is achieved through an innovative system that allows the consultation of various types of sources of information using the native language of each and obtaining the results finla in XML format. In addition, noted that the structure of XML result defined by the user, and may even include data from other sources.

Author: FERNÁNDEZ BERTOA MANUEL.
Year: 2005.
University: MÁLAGA [www.uma.es].
Place of defense: E.T.S.I. INFORMÁTICA.
Place of preparation: E.T.S.I. INFORMÁTICA.

Summary: The last decade has marked the first real attempt to turn development into a software engineering using the concepts of Software Development Based on Components (DSBC) and Component COTS (Commercial Of-The-Shelf) whose goal is to create high-quality items they can be assembled to build a functional system. One of the critical processes within the DSBC is the selection of components that meet the requirements of functionality defined by the user, will be part of the final software product. However, there is a lack of quality models and measures to assist in assessing the quality of software components during this selection process. This thesis presents a set of measures for valuing a characteristic quality of software components, usability. Based on a model of quality for specific components (COTS-QM) and the study of the information provided by vendors of COTS components, formally defined a set of measures to the usability of the software components. In addition, it conducts empirical validation of these measures by developing a family whose experiments and analysis process is described in memory. As a final result will propose a set of indicators for readability, aprendibilidad and operability of software components that are validated and can be calculated in an objective and automatically through a computer analysis of documentation and inspection of software components.

Author: HERNÁNDEZ ACOSTA LUIS MIGUEL.
Year: 2005.
University: LAS PALMAS DE GRAN CANARIA [www.ulpgc.es].
Place of defense: DEPARTAMENTO DE INGENIERÍA ELECTRÓNICA Y AUTOMÁTICA.
Place of preparation: E.T.S. DE INGENIEROS DE TELECOMUNICACIÓN.

Summary: The development of reusable components that can be interconnected and facilitates the development of distributed applications. In this thesis applies a methodology for automating the Robust Application Development and Cooperatives (MADARC). This methodology facilitates the construction of robust software from robust components not available. A prerequisite for applying MADARC is to provide a communication service that understands process group. Although we can use JGroups to implement MADARC, the service closed groups that offer limited certain aspects of design as messages radiated reach all processes of the group. As a result, an increase in customers at a low application performance of the same. In this thesis developed a component that using JGroups, offers an open interface groups and reduces the number of messages. Apart from closed groups, JGroups offers no support for communication Na-1 or NaM. However, a group of processes may require communication with other processes or groups of processes of implementation. In this thesis were two components that provide support for communication Na-1 and NaM with JGroups. The passage of state is a key requirement for high availability applications. In this context usually requires reconfiguration of a group of processes without interrupting the service provided. It has developed a component used JGroups, provides support for cooperative transfer of state. That is, each group member can send part of his state to a new member. MADARC and components described above have been implemented and JDBC RMI to develop an extension of these APls Java that facilitate the construction of robust applications through resting. The new APls offer the same interface but also provide support for the following features: communication Na-1 and NaM, incorporating dynamic new servers to a group of servers and sharing side in the transfer of state. Finally, the new components developed in this thesis has been reused to write a new module for fault detection. This module has been integrated into the stack of protocols JGroups as a new layer.

Author: FERNANDEZ LANVIN DANIEL.
Year: 2006.
University: OVIEDO [www.uniovi.es].
Place of defense: E.U. INGENIERIA TECNICA INFORMATICA.
Place of preparation: DEPARTAMENTO DE INFORMATICA.

Summary: This paper discusses the mechanism that modern object-oriented languages available to the developer for the maintenance of the system consistency. The strength of software is defined as the degree to which a system or component can function properly in the presence of invalid entries of information or environmental conditions with high stress levels. The way to deal with or prevent these cisrcunstancias in object-oriented languages is usually via a mechanism for treatment of exceptions. The treatment of exceptions is an effective tool to manage the errors caused by a problem in the system, but there are certain scenarios where it is inefficient for the recovery of their consistency. This paper conducts a study of the technical and technological resources aimed at maintaining the existing system consistency before the occurrence of an error, especially when the mechanism for treatment of exceptions is not enough for recovery. Based on the conclusions reached after an assessment, a proposed extension semantics led by annotations and applicable to the object oriented languages most popular. The extension complements the treatment of exceptions to prevent the implementation of certain tasks can be automated, saving development effort and reducing the complexity of the source end. In order to test its applicability, the tool has been developed pilot Ariadne, which processes source enriched and, based on the notes associated with attributes and methods, automatically generates the code necessary for the instantiation and management reconstructores, specialized components in restoring the consistency of a specific part of the model. After applying the tool on a set of scenarios model has been proven benefits of the solution designed, opening from the same number of possible lines of work bearing the recoverability proposed mechanism to more ambitious levels of coverage. Finally, the tool has been implemented on PREMOSGAMES, a simulation tool of higher plants developed in the framework of a research project confirm the actual whose characteristics such as the type of application goal of reconstructores.

Author: OCHOA CLAUDIO J. G..
Year: 2006.
University: POLITÉCNICA DE MADRID [www.upm.es].
Place of defense: FACULTAD DE INFORMATICA.
Place of preparation: FACLUTAD DE INFORMATICA.

Summary: The main contribution of this dissertation is the development of the Rating Partial Policontrolada, a novel environment for the evaluation of programs logical part, which is policontrolado in the sense that I was able to take into account sets rules for global and local control, rather employing a unique combination of default (as is the case with traditional partial evaluation). This environment is more flexible than existing approaches, as it allows difererentes control rules to different local and global patterns of call. Thus, it is possible to obtain specialized programs that can not be generated using evaluation parcila traditional. This environment is aware of the resources in the sense that each of the solutions obtained through evaluation partial policontrolada is valued using functions conform. This environment assessment partial policontrolada has been implemented in the system CiaoPP and evaluated numerous test programs. The experimental results obtained in many cases better than those specializations generated using traditional partial evaluation. Another of the contributions of this thesis is the presentation of a unified vision of the elimination of polivarianza superfluous in the assessments and the partial abstract multiple expertise, through the use of a step-minimization grouping equivalent versions of predicates.

Author: SILVA GALIANA JOSEP FRANCESC.
Year: 2006.
University: POLITÉCNICA DE VALENCIA [www.upv.es].
Place of defense: Dep. Sistemas Informaticos y Computacion..
Place of preparation: Universidad Politécnica de Valencia.

Summary: The purification programs can be an arduous task. This is mainly due to the errors may be evident at first glance, or may be hidden in the deeper areas and hidden from our programs. However, surprisingly, is one of the purification processes software that has been least addressed by the scientific community-in fact, the same techniques that were used purification twenty years ago, continue to be used at present. The situation is no different in the context of declarative languages. Or is it even worse, because these languages often present additional difficulties in the depurarlos because, for example, the evaluation mechanism lazy. This thesis examines the current methods for purification of declarative languages and develop new methods and techniques based on the calculation of costs, fragmentation of algorithmic programs and cleansing. Briefly, the main contributions of the thesis are: - The definition of a scheme of calculating costs for programs that functional logic is based on the use of cost centers and to measure different types of costs symbolic. The development of a new technique of fragmentation dynamics based program redex trails, its application to the purification and their adaptation for the specialty program declarative multi-paradigma modern. - The introduction of a new scheme that combines algorithmic purification cleansing algorithmic conventional fragmented programs. - The definition of three new algorithmic strategies for debugging. - The development of a comparative study and classification techniques fragmentation of programs and strategies purification algorithmic.

Author: BLANCO VEGA RICARDO.
Year: 2006.
University: POLITÉCNICA DE VALENCIA [www.upv.es].
Place of defense: Dep. Sistemas Informaticos y Computacion.
Place of preparation: Universidad Politécnica de Valencia.

Summary: This thesis addresses these problems through the Mimetic method, which can be seen as a general way of extracting an explanation (in the form of decision trees or systems rules) of any existing model without considering any internal mechanism of the same (this is considering a black box). The technique can be seen as a mimetic "re-aprendizaje", which generates data, which labeled by the existing model, used to learn a "new" model that mimics or "mimics" the original. First, theoretically studied the behavior of the method used as a tool Mimetic learning curves, we analyze the effect of using or not the joint training of the original model, known scenarios where the technique is beneficial, depending on the accuracy of Mimetic model, etc. .. Secondly, we conducted a pilot evaluation of the technique using datasets from the repository UCI and considering two cases overall study: including whether or not the original dataset. Within this pilot evaluation, we study some of the factors that influence the method, for example, the level of pruning, the level of trust, the size of the fabricated data, and so on. As well as their combined influence. The objective of the study of these factors has been to find an adjustment for them to get a compromise between accuracy and understandability, which has helped establish "procedures" for applying the method Mimetic in an optimum way. Thirdly, and most innovative contribution of the thesis, we expanded the frontiers of applications Mimetic the contextualization of models. Contextualisation or adaptation is necessary in the life of every model and aims to prolong the life of it. The context switches can be presented in different ways, such as the emergence of new data that previously did not exist or were not important, the change in format of the data when they appear or disappear or securities class attributes, modification of costs of mistakes, etc. .. Taking as a basis the mimetic technique is a learning process, in this thesis show that any computational learning technique can be used as a general method of reviewing and adjusting models, regardless of how the model was generated and unused data original or distribution of mismos.De horizontally and with the hearing on the applications has been extended, with respect to previous work, the use of the method Mimetic different types of models "black box": multiclasificadores, neural networks, naive bayes, human experts, etc. We sought to improve the technique for applications in areas where it is needed while the understandability and good precisión.Los different results and contributions of the thesis can be applied to different areas: data mining, reverse engineering (eg. As a model existing get a model that mimics the former, which has the characteristics of our interest: greater accuracy, better intelligibility or a balance between the two), revising and updating models, acquisition and management of knowledge, helping to obtaining knowledge systems experts.