Community:SummerOfCode10:Brainstorming

- A syntax analyser, able o build a semantic tree from a sentence, proceeding by substitutions according to a given set of grammar rules.

- Basic grammar rules and dictionnary for the English language (At this point, the system should be able to interpret all possible meanings simple well-formed sentence)

- A correction system, built in two parts. First, completing the syntax analyser so that it can "guess" the closest correctly formed sentences when the provided one doesn't match with the rules, and second by verifying that all built trees, therefore structurally correct, don't contain concordance mistakes, etc.

It would probably not be ready for everyday use by the end of the SoC (due to the complexity of natural languages making it a lot of work to transcribe them into substitution rules for tree building) but the language-independant part of the system should be operational.

Plus, the syntax analyser could be used later as a basis in a translation tool too, when provided with a base of equivalence between words and sentence structures in two languages.

1.csv and ics import/export: Make sure that the import/export feature of calendar in .csv and .ics format works correctly, Presently either it transfers incomplete data during the export or doesn’t handle the errors(if any)efficiently during import and as a result import fails.

2.Improving the parser and error handling: Make the parser more error tolerant. It should display user understandable error messages when a corrupted file is imported or when a file with wrong extension is intended to be imported.

3.HTML import: Add a feature of importing events in the html format from user's computer as well as the remote URL.

Part 1 and 2 would greatly improve import/export in different file formats and 3 would be a nice new feature.

This project implements Anti-Unified Bug(AUBug) for Bugzilla which utilizes the idea of anti-unification of an abstraction of two or more bug reports. The result of the project shows similar bugs by percentage for an original bug analyzed by a hierarchical collection of abstract structures, which identify the commonalities between two or more bug reports and maintain links to the original bug reports.

In order to achieve high accuracy, the AUBug takes three independent levels/parts to identify the similarity based on user given data of 1) stack trace or 2) source code and 3) textual information using Natural Language Processing (NLP).

Anti-unification techniques are generally used in detection similar code, but we applied it to bug report system. In Mozilla's bugzilla, user-generated bugs including stack trace and source code are rare, but it can be useful in developer-based bug system.