nzhiltsov / mocassin Goto Github PK

1. Filter tokens by their POS categories.
2. Use stems of the tokens.

see TODO in CentralPanel ("this is stub algorithm implementation...")

1. Implement the algorithm.
2. Conduct a test evaluation.

After change any query node chosen before, all the child nodes should be 
cleared.

add support of english language for UI

1. Add filter with typing to suggestion boxes.
2. Implement support to show suggestions even in case of typos

1. Integrate LaTeXML (or the 'make -B omdoc' command functionality) to sTeX 
-> OMDoc conversion.
2. Integrate Krextor API calling to OMDoc -> RDF conversion.
3. Add support to update persistent RDF storage model.

1. Make corresponding combobox items bold.
2. Add keyword string support for query tree widget.

It might be Texlipse parser.

See http://code.google.com/docreader/#p=google-guice&s=google-
guice&t=UsersGuide

Use 'bif:contains' relation of Virtuoso when construct the search query
tree and execute sparql queries.

1. Introduce API for using by thirdparty applications
2. Provide Save-or-Update logic while loading RDF into Virtuoso store

1. Add OpenMath individuals for http://omdoc.org/ontology#Symbol
2. Add UI represenation of these individuals to query input

It's necessary to inject wildcards while query generating if predicate or 
object was unchecked.

1. Persist the OMDoc ontology and the RDF storage in database.
2. Persist additional index information (srcRef etc.) in database.

Not all the elements has 'srcRef' attribute. F.e. <CMP> (omdoc:Property). 
It's necessary to receive it from the parent xml element.

see TODO in QueryManagerImpl.java, line 74

Maybe it would be a widget located to the right of displayed search results.

1. This is for support correct ontology elements even in case of typos.
2. Use Levenshtein distance.

Sometimes there were errors of searching after press this button.

1. Implement the algorithm.
2. Conduct a test evaluation.

Use start/end coordinates of structural element for a given reference

1. Add maven dependency to 'mocassin-virtuoso' module
2. Refactor facade implementations

1. Add link on UI that managed to view HTML page with highlighted relevant 
elements (See Google 'Cached' link).
2. Use srcRef to calculate highlighting regions (at the indexing stage, not 
at query one).
3. Show corresponding HTML page (no need to generate any LaTeX or PDF 
files).

Use containment of labels and corresponding references to infer relations 
between their parent nodes.

Add 'distinct' clause to SPARQL query to get rid of repeated search results

1. Choose 'Доказательство'
2. Choose 'доказывает'
3. The only 'Утверждение' item is displayed at this time.
It's necessary to display all of its subclasses.

See 'RDF' link.
Use DESCRIBE sparql command to get RDF/XML data for given URI.

Add Maven build scripts

The name matcher has to give a mapping between nodes and ontology concepts 
exploiting string similarity algorithms.

Some OWL items are available to choose now. Maybe it's worth to leave Dublin 
Core items (title, creator)  with more attractive labels.

See the following error: "error: SQ074: Line 3: SP031: SPARQL: Internal 
error: The length of generated SQL text has exceeded 10000 lines of code"

At the beginning there is single input. It's sufficient to choose class in 
this input and submit query. Leave arbitrary predicate and object of the 
triple.

See trac.kwarc.info/arXMLiv/ 

Add XPath analyzer to retrieve nodes, edges and their contexts (in particular, 
text around the references).

This should be bound to UI localization change event.

1. Group results by the document on UI (see Google solution)
2. Add server-side support.

Use LingPipe library.

move this functionality from OMDocResourceLoaderImpl.java -> 
QueryManagerImpl.java

Organize the query user input with adding corresponding buttons.

Note: Maybe, it makes sense to represent it as a tree with suggestion boxes 
as its nodes.

Use the texlipse parser to build tree from latex document.

Do it the same as Google search results

F.e. for 'usesSymbol' '=' it doesn't make sense to specify last node.

1. Replace source coordinates information with contents of corresponding 
element on UI.
2. Implement the indexing and retrieving that contents.

Add UI page:
 1) to display statistics of current state of the index.
 2) to run indexing process for given local folder.

Add server-side support for this functionality.

1. Manually annotate the LaTeX source files of articles with sTeX.
2. Get access to another collections.

1. fix tests
2. add svn tag
3. upload stable libraries to '/downloads' folder

F.e., 

1. Choose 'Example' concept
2. Choose 'exemplifies' relation

There will be no any available concepts.

1. Create widget for pagination.
2. Implement server-side support for pagination.

1. Create the separate NLP module.
2. Plug in GATE API.
3. Extract the following features: word-based, POS-based, chunk-based.

Instead of filling the document information, retrieve all the data as 
instance of #Document with SPARQL additional sub-query.