BiblioTECA Implementation

Verba Logica Home page | BiblioTECA Home page | BiblioTECA Implementation | Previous | Next

AFCA module

The AFCA module performs the texts' analysis. The following hypothesis underlies AFCA: the informational content of any text in the intended application field is representable as a tree in which nodes are pairs (label,value). In BiblioTECA current implementation values are always (a transformation of) a lexeme sequence in the original text. Now this simple, well-understood structure can be parsed using reasonably standard tools in Natural Language Processing. This obviously does not infer that the parsing process is trivial. Far from that, parsing actual texts is quite a complicated matter. This is particularly true under BiblioTECA constraints: different kind of documents, some pressure on time and memory requirements and a very high level of correctness and robustness,....

AFCA parsing device, the Parser Engine- is designed and implemented using tools in the logic programming tradition, with heavy emphasis in meta-programming and logic grammars. PROLOG was the selected programming language. The interface, DFA, has been designed using Visual Basic.

AFCA contains two modules: DFA and Parser Engine, which communicate through files. This division follows from the necessity of implementing a state-of-the-art user Windows interface, DFA, without losing the advantages inherent to PROLOG and its associated search and inference engines. The following diagram shows the system architecture main features:

Diagram 3

Verba Logica Home page | BiblioTECA Home page | Previous | Next