Yadoda Plugins
Plugins are jar files that include class binary files inside: classes are loaded when the application starts. Yadoda recognises those classes that implement specific core interfaces, and uses them when needed (there are many hooks in the code). As for now, Yadoda supports four different plugin types.
- Metadata
- By default, Yadoda enriches documents with few metadatas (common to all kind of documents). If the type of a document is supported by a Metadata plugin, then the system will use the plugin to automagically infer metadata of the specified document.
- Relation
- Without plugins, the system lets the user create document relationships manually. Relation plugins are used to infer relationships automatically. As an example, they can be used to create relationships between the documents with the same author.
- Category
- Once the user has created the personal ontology, the Category plugins can be used to automatically suggest the destination categories of the new inserted documents.
- Unifier
- Yadoda's duplicate finder method is based on the MD5 checksum algorithm: if two documents have the same checksum, they are equal. This is always the case, of course, but it turns out to be a too strict method. In fact, the comparison failure doesn't mean the two documents are really different: as an example, they could be the same original document and still differ for few bits or even for the file format (one is pdf, the other is html). A Unifier plugin can be implemented to provide smart methods.