Difference between revisions of "Task ideas for Google Code-in"

From Apertium
Jump to navigation Jump to search
Line 350: Line 350:
|title=script to test coverage over wikipedia corpus
|title=script to test coverage over wikipedia corpus
|mentors=Jonathan
|mentors=Jonathan
|description=Write a script (in python or ruby) that in one mode checks out a specified language module to a given directory, compiles it (or updates it if already existant), and then gets the most recently nightly wikipedia archive for that language and runs coverage over it (as much in RAM if possible). In another mode, it compiles the language pair in a ___ that it disposes of after successfully running coverage.
|description=Write a script (in python or ruby) that in one mode checks out a specified language module to a given directory, compiles it (or updates it if already existant), and then gets the most recently nightly wikipedia archive for that language and runs coverage over it (as much in RAM if possible). In another mode, it compiles the language pair in a docker instance that it then disposes of after successfully running coverage.
|tags=python,ruby,wikipedia
|tags=python,ruby,wikipedia
}}</table>
}}</table>

Revision as of 22:25, 3 September 2017

Contents

This is the task ideas page for Google Code-in, here you can find ideas on interesting tasks that will improve your knowledge of Apertium and help you get into the world of open-source development.

The people column lists people who you should get in contact with to request further information. All tasks are 2 hours maximum estimated amount of time that would be spent on the task by an experienced developer, however:

  1. this does not include time taken to install / set up apertium.
  2. this is the time expected to take by an experienced developer, you may find that you spend more time on the task because of the learning curve.

Categories:

  • code: Tasks related to writing or refactoring code
  • documentation: Tasks related to creating/editing documents and helping others learn more
  • research: Tasks related to community management, outreach/marketting, or studying problems and recommending solutions
  • quality: Tasks related to testing and ensuring code is of high quality.
  • interface: Tasks related to user experience research or user interface design and interaction

You can find descriptions of some of the mentors here: List_of_Apertium_mentors.

Task ideas

typetitledescriptiontagsmentorsbgnr?multi?
code Refactor/mege the main "processing" functions of lrx-proc lrx-proc has two modes, "-m" mode and default mode. They are implemented by each their huge function, nearly identical to each other. Refactor the code to remove the redundancy, and run tests on lots of text with several language pairs to ensure no regressions. c++ Fran, Unhammer
code Profile and improve speed of lrx-proc lrx-proc is slower than it should be. There is probably some low-hanging fruit. Try profiling it and implementing an improvement. c++ Fran, Unhammer
research See if you can precompile xpath expressions or xslt stylesheets An XSLT stylesheet is a program for transforming XML trees. An Xpath expression is a way of specifying a node set in an XML tree. Investigate the possibility of pre-compiling either stylesheets or xpath expressions. parsing Fran
research Review literature on linearisation of dependency trees A dependency tree is an intermediate representation of a sentence with no implicit word order. Linearisation is finding the appropriate word order for a dependency tree. Do a survey of the available literature and write up a review. parsing Fran, Schindler
research Manually annotate/Tag text in Apertium format Take some running text, analyse it using an Apertium analyser then manually disambiguate the result. Fran
code Convert Chukchi Nouns to HFST/lexc There is a freely available lexicon of Chukchi, a language spoken in the north-east of Russia. The objective of this task is to convert part of the lexicon covering nouns to lexc format, which is a formalism for specifying concatenative morphology. Fran
code Convert Chukchi Numerals to HFST/lexc There is a freely available lexicon of Chukchi, a language spoken in the north-east of Russia. The objective of this task is to convert part of the lexicon covering nouns to lexc format, which is a formalism for specifying concatenative morphology. Fran
code Convert Chukchi Adjectives to HFST/lexc There is a freely available lexicon of Chukchi, a language spoken in the north-east of Russia. The objective of this task is to convert part of the lexicon covering nouns to lexc format, which is a formalism for specifying concatenative morphology. Fran
interface Make a design for a web-based viewer for parallel treebanks (also for viewing diff annotation for same sentence) dependencies,parallel,web Fran,Jonathan
code Write a script to convert a UD treebank for a given language to a format suitable for training the perceptron tagger
research Train the perceptron tagger for a language The perceptron tagger is a new part-of-speech tagger that was developed for Apertium in the Summer of Code. Take a language from languages and train the tagger for that language. Fran
interface Design an annotation tool for disambiguation like c.f. webanno, corpus.mari-language.org, brat disambiguation,annotation Fran,Jonathan
interface Design an annotation tool for adding dependencies Like c.f. brat dependencies,annotation Fran,Jonathan
code Train lexical selection rules from a large parallel corpus for a language pair Fran
documentation Document how to set up the experiments for weighted transfer rules Fran
code convert UD treebank to apertium tags, use unigram tagger (see #apertium logs 2016-06-22)
code Write a script to extract sentences from CoNLL-U where they have the same tokenisation as Apertium. Fran, wei2912
documentation convert [1] to apertium-style documentation Schindler
code Implement `lt-print --strings` lt-print -s c++ Fran, wei2912
code Implement lt-expand -n Implement an algorithm that prints out a transducer but only follows n cycles. c++ Fran, wei2912
code in-browser globe with apertium languages as points Use d3 globe to make an apertium language/pair viewer (like pairviewer), maybe based on this or this or this. This file contains coordinates of Apertium languages. js,html,maps Jonathan, kvld
code Write a program to detect contexts where a path in a compiled transducer begins with a whitespace c++
code Make the lt-comp compiler print a warning when a path begins with a whitespace. Common mistake in dix files is to have some bad whitespace at places, this needs to be aqutomatically detected in the compilation tool and warning to user issued. c++
apertium-mar-hin: make the TL morph for any part of speech less daft Some morph in Marathi or Hindi are currently daft. morphology vin-ivar
add indic scripts/formal latin transliterations Translitteration is a ways to write stuffs in different scripts. Currently some indic scrpts are done only to some WX transliterator python vin-ivar
code apertium-hin: more consistency with apertium-mar for verbs Verbs in Marath and Hindi are incosistently. morphology vin-ivar
code apertium-mar: replace cases with postpositions Marathi cases are postpositions morphology vin-ivar
code apertium-mar: fix modals and quasi-modals Modals in Marathi need fixing morphology vin-ivar
code refactor x file in apy Reorganise apy code to be more readable, maintainable and so forth. Putti
documentation add docstrings to x file in apy docstrings are a way to document python code that can be generated into documentation on the web or in python. See following PEPs in python.org Putti, vin-ivar
quality write 10 unit tests for apy Putti, Unhammer, (sushain?)
code add 1 transfer rule Transfer rules are parts of translation process dealing with re-arranging, adding and deleting words. See also Short introduction to transfer Fran, vin-ivar, zfe, kvld
code add 50 entries to a bidix Bilingual dictionary (bidix) contains word-to-word translations between languages, e.g. cat-chat or cat-Katze in English to French or German respectively. Add 50 of such word-translations to languages you know. Fran, vin-ivar, zfe, kvld, Schindler
code write 10 lexical selection rules Write 10 lexical selection rules for a pair already set up with lexical selection Fran, vin-ivar, zfe, Unhammer
code write 10 constraint grammar rules Constraint grammar is a rule-based approach of selecting linguistic readings from ambiguous cases, to improve translation quality etc. See introduction CG here: Fran, vin-ivar, zfe, kvld, Unhammer
research Document resources for a language Document resources for a language without resources already documented on the wiki. read more... Jonathan, vin-ivar, zfe, Schindler X X
research Write a contrastive grammar Document 6 differences between two (preferably related) languages and where they would need to be addressed (morph analysis, transfer, etc). Use a grammar book/resource for inspiration. Each difference should have no fewer than 3 examples. Put your work on the Apertium wiki under Language1_and_Language2/Contrastive_grammar. See Farsi_and_English/Pending_tests for an example of a contrastive grammar that a previous GCI student made. vin-ivar, Jonathan, Fran, zfe, Schindler X X
research apertium-hun: match existing apertium-hun paradigms with morphdb.hu Morphdb.hu is another implementation of Hungarian morphology, that has a large lexicon. In order to convert it to apertium format, the classification of the words needs to be mapped to one used in apertium. hun,dix Flammie
code apertium-hun: convert hunmorph.db into apertium one of: See prerequisite task above. Flammie
code apertium-fin-eng: go through lexicon for potential rubbish words) Apertium's Finnish–English dictionary has been converted from projects, like Finnwordnet, that hae a lot of pairs unsuitable for MT, find and delete them from the file. fin,dix Flammie
code apertium-fin-eng: add words from apertium-fin-eng to apertium-eng grep for English words in apertium-fin-eng.fin-eng.dix and classify them according to paradgims. See also: Apertium English) eng,dix Flammie
code apertium-apy: add i/o formats) Currently APY web queries get responses in ad hoc json format. Research and implement interoperabilities with further formats, such as: apy Flammie
code apertium-apy: write metadata about apertium language pairs CMDI format that can be deployed for CLARIN stuffs apy Flammie
code apertium-apy: make more parts of apertium-pipeline on web apertium.org has a web service interface for getting translations or morphological analyses. This should be extended for other functions of apertium as well. more information: Apertium Apy. apy Flammie
code Finish suggest-a-word feature so it can be deployed to apertium.org There exists a version from last GSOC of apertium.org translator where user can suggest fixes to unknown word translations among other things, but this is not deployed to server. apy Flammie
code Further developments to suggest a word Currently suggested words may be added to wiki by a service, it would make sense to also have e.g. chance to login and get attributed as contributor, as well as other stuff ) apy Flammie
code Fix ordering of dependencies in CG matxin format Fran
code CG syntax highlighting plugin for a text editor Write a syntax file for your favourite text editor that provides fancy syntax highlighting for Constraint Grammar vin-ivar, Unhammer, (Flammie?)
code Package apertium-lint to install to a prefix apertium-lint currently installs with pip, modify that to allow passing a flag for installing it to a prefix vin-ivar
quality Fix a bug in Apertium html-tools Fix a currently open issue with html-tools in consultation with your mentor. multi,html,js,html-tools Unhammer, Jonathan, Kira X
quality Fix a bug in Apertium APy Fix a currently open issue with APy in consultation with your mentor. multi,python,apy Unhammer, Jonathan, Kira X
code Script to get resources from GF Write a script to scrape words from one particular paradigm in GF and make it usable in Apertium. vin-ivar
code Create a list of text editors compatible with different scripts Create a list of ten text editors and document their status with representing human text (Latin), RTL text (Arabic), combining characters (Devanagari), etc. Document any bugs with eg. copy/paste and tab indentation. vin-ivar
code Write a script to strip apertium morphological information from CONLL-U files Write a script to strip apertium morphological information from CONLL-U files so the dependency trees can be rendered okay by the online tools. vin-ivar
research Investigate FST backends for Swype-type input Investigate what options exist for implementing an FST (of the sort used in Apertium spell checking) for auto-correction into an existing open source Swype-type input method on Android. You don't need to do any coding, but you should determine what would need to be done to add an FST backend into the software. Write up your findings on the Apertium wiki. spelling,android Jonathan
code Fix a memory leak in matxin-transfer The matxin-transfer program is a component of the Matxin MT system, a sister system to Apertium. Run valgrind on the code and find and fix a memory leak. There may be serveral. c++ Fran
code Write a tool helping to test a bidix coherence This tool will generate a file with each lema of the main categories (at least nouns, adjectives ans verbs) found in a bidix. Then this file will be translated to the second language and back to the first one. Looking for changes will allow to detect transfer problems and changes of meaning. Bech
quality fix any begiak issue Fix any open issue for begiak (Apertium's IRC bot), to be chosen in consultation with your mentor. python,irc Jonathan, sushain, wei2912 X
quality merge phenny upstream into begiak Merge upstream patches etc. into begiak (Apertium's IRC bot). git,irc Jonathan, sushain, Unhammer, wei2912
quality open a pull request for merging begiak modules into upstream Open a pull request to merge features from begiak (Apertium's IRC bot) into upstream. git,irc Jonathan, sushain, Unhammer, wei2912
code begiak interface to Apertium's web API Write a module for begiak (Apertium's IRC bot) that provides access to at least one feature of APy (Apertium's web API). You may want to base the code off begiak's Apertium translation module (which may not be in 100% working order...). irc,apy Jonathan, sushain, Unhammer, wei2912 X
research tesseract interface for apertium languages Find out what it would take to integrate apertium or voikkospell into tesseract. Document thoroughly available options on the wiki. spelling,ocr Jonathan
interface Abstract the formatting for the Html-tools interface. The interface for html-tools (Apertium's website framework) should be easily customisable so that people can make it look how they want. The task is to abstract the formatting and make one or more new stylesheets to change the appearance. This is basically making a way of "skinning" the interface. css,html-tools Jonathan,sushain
quality improvements to lexc plugin for vim A vim syntax definition file for lexc is presented on the following wiki page: Apertium-specific conventions for lexc#Syntax highlighting in vim. This plugin works, but it has some issues: (1) comments on LEXICON lines are not highlighted as comments, (2) editing lines with comments (or similar) can be really slow, (3) the lexicon a form points at is not highlighted distinctly from the form (e.g., in the line «асқабақ:асқабақ N1 ; ! "pumpkin"», N1 should be highlighted somehow). Modify or rewrite the plugin to fix these issues. vim Jonathan
code Write a transliteration plugin for mediawiki Write a mediawiki plugin similar in functionality (and perhaps implementation) to the way the Kazakh-language wikipedia's orthography changing system works (documented by a previous GCI student). It should be able to be directed to use any arbitrary mode from an apertium mode file installed in a pre-specified path on a server. php Jonathan
documentation add comments to .dix file symbol definitions dix Schindler
documentation find symbols that aren't on the list of symbols page Go through symbol definitions in Apertium dictionaries in svn (.lexc and .dix format), and document any symbols you don't find on the List of symbols page. This task is fulfilled by adding at least one class of related symbols (e.g., xyz_*) or one major symbol (e.g., abc), along with notes about what it means. wiki,lexc,dix Schindler
code conllu parser and searching Write a script (preferably in python3) that will parse files in conllu format, and perform basic searches, such as "find a node that has an nsubj relation to another node that has a noun POS" or "find all nodes with a cop label and a past feature" python,dependencies Jonathan, Fran
code group and count possible lemmas output by guesser Currently a "guesser" version of Apertium transducers can output a list of possible analyses for unknown forms. Develop a new pipleine, preferably with shell scripts or python, that uses a guesser on all unknown forms in a corpus, and takes the list of all possible analyses, and output a hit count of the most common combinations of lemma and POS tag. guesser,transducers,shellscripts Jonathan, Fran
code vim mode/tools for annotating dependency corpora in CG3 format includes formatting, syntax highlighting, navigation, adding/removing nodes, updating node numbers, etc. vim,dependencies,CG3 Jonathan, Fran
code vim mode/tools for annotating dependency corpora in CoNLL-U format includes formatting, syntax highlighting, navigation, adding/removing nodes, updating node numbers, etc. vim,dependencies,conllu Jonathan, Fran
quality figure out one-to-many bug in the lsx module C++,transducers,lsx Jonathan, Fran
code add an option for reverse compiling to the lsx module this should be simple as it can just leverage the existing lttoolbox options for left-right / right-left compiling C++,transducers,lsx Jonathan, Fran
quality remove extraneous functions from lsx-comp and clean up the code C++,transducers,lsx Jonathan, Fran
quality remove extraneous functions from lsx-proc and clean up the code C++,transducers,lsx Jonathan, Fran
code script to test coverage over wikipedia corpus Write a script (in python or ruby) that in one mode checks out a specified language module to a given directory, compiles it (or updates it if already existant), and then gets the most recently nightly wikipedia archive for that language and runs coverage over it (as much in RAM if possible). In another mode, it compiles the language pair in a docker instance that it then disposes of after successfully running coverage. python,ruby,wikipedia Jonathan