CORDIS proporciona enlaces a los documentos públicos y las publicaciones de los proyectos de los programas marco HORIZONTE.
Los enlaces a los documentos y las publicaciones de los proyectos del Séptimo Programa Marco, así como los enlaces a algunos tipos de resultados específicos, como conjuntos de datos y «software», se obtienen dinámicamente de OpenAIRE .
Resultado final
IRAIA is a cross-provider service platform that enables information providers to establish their individual information service based on individual as well as thematically corporate information architectures. The most outstanding element of IRAIA's human-computer interaction is concept hierarchies that provide the users always with the necessary orientation while exploring the information space.
Unlike in traditional search engines, in IRAIA users search or navigate only by selecting entries (terms or phrases) in concept hierarchies. Defining a query means simply clicking on entries. There are the following products/product versions of IRAIA available for immediate application and for further development in terms of improvement:
- Basic version: The basic version of IRAIA will be available for free;
- Advanced version: The major business opportunities for the IRAIA software provider lie in development and sale of advanced and special tools. The freely available basic tools help branding their technical platform by fostering the transition from early adopters to non-technical mainstream users. Like "a rising tide lifts all ships", it can be expected that the added-value of these free basic tools will increase the sales of the advanced and powerful versions of IRAIA. These versions cover features required by more advanced services and new innovations that will be integrated into the system after the project's lifetime. With each new significant set of innovative features the former system will be "relegated" to freeware or open source and a new advanced tool goes on sale;
- Individual version: Further economic benefits for the software developer arise from the installation and adaptation of provider-dependent specifics (individual tools). Once implemented, IRAIA can be adapted and customised to individual requirements. The open interfaces of the IRAIA components protect the core technology from customisations without restricting the flexibility of the service provider. IRAIA implementations are not at risk of becoming "orphan installations" that are cut off from future upgrades in an environment of rapid change. Thus IRAIA easily integrates with in-house and legacy systems.
IRAIA put a strong emphasis on the integration of different taxonomies and the development of metadata. The special approach followed in this work is the pre-structuring of the keywords. It was presumed that the structures do not only guide the user with more precision and less time to the information he needs but that these structures also make it possible to easily handle different languages. The user can start his retrieval out of the structures in his native language and reach texts and figures in his language and at the same time in all the other incorporated tongues. The system developed here represents only one model out all the possible fields of human interest. It was devoted to economics because the design of such a system must be based on a lot of expertise within the field in focus and among the partners appear two institutes for economic research, the DIW, Berlin and the IFO-Institute, Munich. In the beginning the structures practically were derived from the structures of economic time series. These in most cases are described by three dimensions: Theme, Sector and Geographic Region e.g. Production/Textiles/France.
Discovering sequential relationships among time series is important to many application domains. In data mining applications, it is often necessary to search within a series database for time series that matches a pre-specified query series. This primitive is needed, for example, for prediction and clustering purposes. Clustering of time series data contributes to the problem of inducing and forming categories (classes) of events. For example the problem of finding trends, seasons and cycles in a sole time series may be approached by finding similar parts (or, segments) of the series itself. Moreover, the identification of time series coherences could be also approached by the identification of similar ordered sub-sequences between the time series. In a cross-series analysis clustering helps to find indications of one series having an economic impact on a further one. Time series express economic phenomena only by virtue of their numeric content. A phenomenon in terms of an economic variable having an impact on a further one can barely be found on the basis of the textual information attached to series. The information on the impact is contained in significant sequences of the numeric values alone.
The tool developed by Forth help to discover this coherence information buried in numeric data. The main concern of the undertaken work was to specify ways to integrate time series coherences with coherent text-reference collections, both stored in the IRAIA database server. The final prototype (components for time series similarity assessment and clustering) is built in the Java environment in order to achieve high degree of interoperability. The decision made within the IRAIA consortium was that the system is to reside at the server-side of the IRAIA system. A scenario of utilising the MTSD module has as follows:
- The MTSD module runs over the (potentially) recalled time series, and their coherence (similarity with) other time series in the IRAIA database is assessed;
- Experts in the field evaluate the final outcome, and the validated similar time series associations are recorded in the database (the potential of recording clusters of similar/ coherent collections of time series is also possible).
QUB's visual user interfaces express domain-related taxonomies, which are based on IRAIA's concept hierarchies. Visualising information and data stored in databases or in unstructured or semi-structured repositories is important for the following reasons:
- It allows the user to have some idea before submitting a query as to what type of outcome is possible. Hence visualisation is used to summarise the contents of the database or data collection.
- The user's information requirements are often fuzzily defined at the outset of the information search. Hence visualisation is used to help the user in their information navigation, by signalling related items, by showing relative density of information, and by inducing a (possibly fuzzy) categorisation on the information space.
- Visualisation can therefore help the user before the user interacts with the information space, and during this interaction. It is a natural enough progression that the visualisation becomes the user interface.
The Kohonen self-organising map applied in IRAIA is an unsupervised neural network that provides cartographic visualisation of the results. Implementation was based on a combination of Javascript, CSS and DOM. The server takes the computational load. Customisation is provided, in the display delivered to the user. A number of querying sessions can be carried out simultaneously by the user. These technologies and DHTML were used in our visualisation implementation because they are very appropriate for creating an effective, interactive, dynamic environment with Kohonen maps.
Organising documents in order to make their access more efficient is a crucial problem when dealing with large collections. A first type of methods used in information retrieval systems to organise documents consists in building indexes that make a link between descriptors and document contents. When a free text query is submitted to the system, it is analysed in order to extract the descriptors that match it and the associated documents are retrieved.
Another technique consists in grouping documents according to their content so that documents that deal with the same topics are grouped together. This technique is much more powerful if groups are made according to the users' needs. Categorisation addresses this task. It consists in associating documents with predefined categories that can be organised hierarchically. Document access is then achieved by browsing the corresponding categories. When selecting a node, the user can access the associated documents.
In IRAIA, texts are categorised according to taxonomies that describe the knowledge of the economic domain. In IRAIA, a taxonomy corresponds to a formal hierarchy of terms organised into classes and sub-classes along a generic dimension and connected by synonymy relationships. Browsing these semantic structures leads to the selection of relevant concepts and the definition of a query composed of selected concepts and their description. Expression of a query becomes an instantiation of several concepts.
The association of textual documents to concept hierarchies is based on an IRAIA's "vector voting method". The rationale of this method is that the more terms from a category occur in a text, the stronger is the link between the text and that concept. In fact, the principle is more complex as the terms corresponding to a category are not only terms from the concept itself, but also those associated with the category ancestors and semantically linked terms (eventually resulting from a learning phase). In addition, the weighting function takes into account different term frequencies so that the voting weight is proportional to the frequency of the term in the document and inverse proportional to the frequency of the term in the hierarchy.
Qualitatively enhanced access to data collections requires a more meaningful interface, which endows the users with personal agents that are easily adaptable to the users' individual retrieval preferences. With this feature IRAIA contributes to the development of pragmatic web technologies, where through end-user programming the users create important parts of their own interactive access devices. Tracking individual needs of users leads to personal digital assistance that appears in the system's interaction mode as the user's personal software agent.
The main objective of personal assistance as envisaged in IRAIA is enabling the system to give the users recommendations that are derived from best practice in information search processes. Best practice reflects the knowledge about how users perform successful retrieval activities and attempt to define the best ways of searching large document collections. Best practices are developed on an individual as well as on a community base.
The personal agent draws on IRAIA's model of context-oriented retrieval and observes, records and analyses the users' search and navigation by applying concepts of the respective taxonomies. In IRAIA the personal agent is realised as an add-on to a Java application representing the search interface (i.e. the client component of IRAIA). It derives recommendations from the search and navigation history of the individual user or the user community.
Personal assistance starts with observing the user's selections of entries from the concept hierarchies, i.e. with capturing query profiles. An actual profile then is compared with already stored ones and recommendations are derived if the agent can recognise a certain similarity between the actual profile and one of those regarded as a representative of a recommendable search or navigation strategy. The recommendation refers to further selections of concepts suitable in an actual retrieval situation.
Buscando datos de OpenAIRE...
Se ha producido un error en la búsqueda de datos de OpenAIRE
No hay resultados disponibles