This tool has been developed to take advantage of the underused mouse middle-button. The idea is to extend the middle-button click action adding new Semantic capabilities. These capabilities are illustrated in a contextual popup menu. This menu allows user to annotate text of any opened file, stablish relationships between various documents and insert already made annotations into an opened document. In order to help and restrict user annotations, menu only shows the classes, properties and relationships derived from domain ontology. This way, SeMouse becomes an intuitive and simple tool for populating ontologies. SeMouse has been designed and developed to be editor and file independent, and this goal has been achieved descending to the Operative System level. However, the drawback of that is that user only can make shallow annotations, i.e. only text is anotated. But, this drawback is thwarted by the advantages provided such as related files traversal, reutilization of annotated text and semantic search of documents.

Main Features Summary

• Takes advantage of the underused mouse middle-button: Due to the fact there are very few applications that use the middle button we decided to averride its normal behaviour (if any) to provide semantic capabilities.
• Application independent annotation tool: SeMouse has been developed at Operating System level which provides a complete application independece, this way it is possible to annotate, author and associate any kind of documents.
• Ontology Domain-based menu structure: once the ontology is loaded into the Knowledge Database (KDB), the menus are driven by this ontology, in the sense that displayed classes, properties and relations fit to the ontological model.
• Authoring tool: if the editor behind the mouse pointer allows edition, it is possible to append information of other documents associated with the current one. This way, user saves the time required to find and open the document and copy and paste the desired information.
• Allows associating documents: with SeMouse it is easy to create a network of documents semantically related, following the associations defined in the ontology.

 This section shows a typical user usage of seMouse tool. Here, usual steps are depicted, but some of them may be skipped depending on user's needs. Before entering into the matter, and in order to be better understood, a sample ontology is provided. In this case University-project domain is used. In the figure 1 a UML representation is shown of a possible representation of that domain. The red square that appears in the middle of the figure, permits to center tha attention in the portion of the ontology which will be used during the sample.

 The typical usage of the tool can be split in the following 4 steps:

1. File Classification: With the file to be annotate opened, the first time the user clicks the mouse middle button a menu is displayed to let user classify the document according to the classes specified by the domain ontology. Figure 2 shows an opened PDF file and the classification menu. In this case, the opened document is a Deliverable.

   

   

2. File Annotation: Once user has said that this file is a Deliverable, he/she can start with the annotation process. That is achieved selecting the fragment of text which he/she wants to give a meaning. For example, he/she could highlight the title and afterwards click the middle button, then a menu pops up with the properties of the previously selected class. In this case, the highlighted text maches with the title property so user only has to select that option (figure 3).

   

   

   • Checking annotations: One way to check if the annotations has been done correctly is querying metadata repository, but that would be very cumbersone for the user so, we provide a simpler way to achieve that, user only has to click middle button without selecting any text. The new menu will show the properties of this file and the values of each of them. Figure 4 shows this.

     

     

3. Associating documents: This is probably one of the most interesting functionalities provided by seMouse. Anyone can create an association between 2 documents following the relationships described in the ontology. The menu shown in the figure 5, presents how an association between a project document and a deliverable is created. In the submenu related to delivers association appears every instance whose class is the same than the range class of the delivers relationship. Associations can be removed at any time like it is shown in the figure 6, just clicking in the remove association menu option.

   

   

   

   

4. Authoring documents: Once association between documents has been stablished, user may take advantage of this fact to include the information of the other endpoint of the association into the already opened document. Each associated instance has a submenu that allows traverse the properties and values of that instance. Selecting a value causes the value to be pasted in the current position of the cursor. We think this way user can take advantage of the prevously made annotations, and save the time that a task such that could involve if it is done manually. Figure 7 exemplifies previously said.

   

   

Althought this seems a really simple tool, the work behind scenes was quite hard, because we had to resolve the challenges that operating system raised every time. The most difficult issues we had to face to was the opened file path retrieval and the text capture from external applications.

Drawbacks

• Plain annotation of metadata --> only plain text: due to seMouse's decoupled nature it can not make annotations so complex as other annotation tools do. Those tools are usually focused on a very few kinds of documents or are completely integrated into a particular editor. That provides a better control of the annotations, giving also the chance of store annotations within the document, however, this tight integration hinders the spread of that tools to other types of documents.
• Rigid menu structure --> no extensible: it may be interesting allow users to personalize their menus, for example, to extend the operations he/she wants to perform. A possible way of doing that is adding the ability of adding plug-ins which take advantage of the information retrieved by seMouse.

• Application independent: this allows annotating any kind of document wihout changing the annotation tool.
• Domain independent: it is possible to reuse any ontology.
• Simple ontology population: user can add new instances to the KDB without knowing the details of the ontology which is loaded.
• Easy to create relationships: the user does not have to learn which instance should use in each association, the given menu constraints the selection to the only suitable ones.
• Easy authoring: once the association is stablished, authoring is so easy as selecting the desired instance's property value and click, the value automatically is pasted at the current cursor position.

Future Work

• Add operations to classes: current version only supports authoring and annotation, but other operations like browsing or visualization could be interesting in some contexts.
• Allow users to create the ontology at the same time he/she annotates: the rigid structure of displayed menus driven by a  priori developed ontology may discourage users from adopting  this tool, because it limits the freedom of describing what they want to say about a resource. A possible alternative to this, is to provide to the users the possibility of creating their own ontologies at the same time they annotate the information, because at annotation time users are aware of the selected texts semantics. 
• Use Web Services to retrieve information about documents: this point could be considered such as one of the operations that could enrich the tool. In this case, the domain should be constrained to anyone which could have documents published in the Web. What this operations should do is once that title of a paper has been selected by the user, query a web service to retrieve the rest of information like authors, conference, etc...
• Publish annotated information in the web (e.g. into a Blog): another operation, which could be interesting for share the annotations between the members of a research group, for example.
• Annotation of special information items such as email messages or images: messages need a special treatement because they are not isolated files, but members of a bigger structure and images does not contain any text that could be captured.

• Oscar Díaz
• Jon Iturrioz
• Sergio F. Anzuola

1. Oscar D¡az, Jon Iturrioz and Sergio F. Anzuola, Authoring and annotation of desktop files in seMouse in 1st Semantic Desktop Workshop, Pages 233-237, Galway (Ireland) November 2005
   http://www.semanticdesktop.org/SemanticDesktopWS2005/final/8_diaz_semouse_poster.pdf