Towards Semantically Structuring GitHub

Abstract

With the recent increase of open-source projects, tools have emerged to enable developers collaborating. Among these, git has received lots of attention and various on-line platforms have been created around this tool, hosting millions of projects. Recently, some of these platforms opened APIs to allow users questioning their public databases of open-source projects. Despite of the common protocol core, there are for now no common structures someone could use to link those sources of information. To tackle this, we propose the SemanGit ontology, the rst ontology dedicated to the git protocol, which also describes GitHub's features to show how it is extensible to encompass more git-based data sources.

-

Introduction Open-Source technology is, once accepted as bene cial, subject to improvement attempts under the premises of commercial marketing, ideological beliefs or just feasibility of implementation. One of the most popular examples is Linux, with its variety of ecosystems, ranging from commercial server distribution to opensource desktop implementations. While not considered as heterogeneous in its implementations, for now, the git protocol [ 10 ] faces the same development.

Developed in 2005 as a distributed version-control system, git [ 10 ] is tracking changes in a le system while providing several properties such as data integrity or support for distributed and non-linear work ows. Since the le system represented by git can be distributed, developers embed their changes into a local git repository and later \push" their contributions to an online repository so that collaborators can then have access to their modi cations and contributions. Quickly, the git protocol has evolved to provide more and more features dedicated to large open-source communities and projects. In recent years, more and more platforms emerged using the git protocol to provide their users with a version-control system, with every one of them using additional features to provide a better user experience, faster distribution or improved maintainability.

Nowadays, GitHub seems to be the most popular platform [ 1 ] for the external usage of git, but other providers like GitLab achieve more and more popularity [ 1,2 ]. As all these systems share a common infrastructure, based on git and principles adopted from social networks, it is possible to agglomerate the different approaches in a uni ed model and to merge these providers' data into a common source of information. In this study, we present the SemanGit ontology, an OWL ontology serving three goals. First, it structures data provided by the git protocol. Second, it models as much information as possible from GitHub. Third, it contains enough abstraction such that similar information from other sources like BitBucket, CloudForge, GitLab, Launchpad can have a consistent representation, using the same parental classes.

The SemanGit ontology can then be used to collect data about developers from di erent domains, interlinking information about publicly available software, its developers and interactions between them with social media and research platforms. While we propose not the rst ontology related to data from git [ 6 ], our approach is a novelty, focusing on the possibilities of automated data extraction and analysis. Indeed De Nies et al. proposed the git2PROV system, tailored to the task of extracting provenance information [ 6 ] as a feature for developers publishing their work.

More generally, we fall within the domain of structuring public data (see e.g. [ 7 ] focusing on describing projects). So far, numerous projects are already providing such datasets with each one tackling a distinct domain. Among this list, we can mention DBpedia [ 3 ] which proposes a linked version of Wikipedia, or also LinkedGeoData [ 4 ] which deals with geographical data. Both are excellent targets for a future interlinking of data extracted from git hosters, to enrich the information about users origin. 2

A git Ontology and its GitHub Extension The git protocol relies on so-called repositories for storing les and tracking data modi cations. A large share of online git repository providers add features of their own that are not part of the git protocol, such as social features. To create an extensible ontology, we need to implement a strict distinction between what is part of the git protocol and what is provider speci c. As an example, according to the git protocol, the author of a commit is a pair \Name <email>" whereas on GitHub an author, i.e. a user, is much more complex. It has additional attributes such as a creation date, an avatar, a location and even social-featured ones such as e.g. an associated website. The part of the SemanGit ontology covering the git protocol features only the data that strictly belongs to the protocol. The classes in this section mostly form the basis from which platform-speci c classes inherit, see Figure 1 for an example. This protocol-related part is rather small and comprises of merely four classes: users, projects (i.e. repositories), commits and pull requests, the user class storing only an email address. The projects refer to a URL, a time stamp of its creation and the commits that were submitted to it. The other two classes are slightly more complex as commits have a hierarchical structure in themselves and pull requests are requests to accept a cross-branch commit, possibly coming from another project. Seeing that all extensions of the git protocol are still required to provide the base functionality, we have chosen a hierarchical approach for the SemanGit ontology, letting extensions inherit from protocol-conform classes all properties they are expected string string integer dateTime Literal

github user lng githubggituihthsueubrb uduseseleerrtelgodcicrtheauatibtoendusaetrid githubgiuthsuebr ccooumnptgragyintihtyhgcuuiobthbdueubusseeurrsecsrittyalatet github login github user

github user is org boolean github user fake

boolean Subclass of commentable github fol ower

Subclass of

github issue assignee github issue reporter github issue string string

Literal github issue event actor github project joining user github has owner github fol ow. . github fol ows fol owable

Subclass of github issue pul request

Subclass of

Literal string commit sha user email commit ccoommmmititedaubthyor commit has parent commit created at dateTime comment author Subclass of boolean string comment integer comment pos comment for comment line comment created at comment body commit integer dateTime commit message string

Subclass ostfring

repository has commit

Subclass of github commit

user

Literal github issue event for

github issue. . github issue created at github issue id github issue event created at

github issue event action github issue event action specific sha dateTime string string dateTime integer github issue label used by github issue project github projec. .

github project join event created at

dateTime github project joined the ontol ogithgubryepo label project thgeithub gprojeict t, tgihthubeproject name github repo . .

string github project description gitlhuabfoyrgkitehduebfroprmrogjesitchtubdeploergtoeijtedhfcutb idproject has mgiiltehssuttrboinngeproject language repo

github repo label name integer github projec. . github project lang

github project la github projec. .

github giDthautbatyppreoject milestone name (external)

string github pul r. .

github pul request action type github pul request action id github pul request action created at dateTime compass additional features from public open-source platforms by considering GitHub. The full SemanGit ontology is publicly available for further community driven development on GitHub1 and on our website2. An interactive visualization can be found on VisualDataWeb3. While focusing on the structure and features of GitHub for the moment, we designed our ontology to be extensible by the information generated by any other host of git based version-control systems. We created this ontology as a starting point to build semantic datasets from various collaborative-platforms. Moreover, we already built one from GitHub: the SemanGit dataset [ 8 ]. Such datasets could then allow innovative perspectives for data analysis if one considers an enrichment of data from other sources like DBpedia [ 3 ], LinkedIn [ 9 ] or linking researchers code development with ScienceGRAPH [ 5 ]. While the SemanGit ontology is currently published at our homepage, we strive for the integration on external hubs such as LOV [ 11 ] in the near future. 1 https://github.com/SemanGit/SemanGit/blob/master/Documentation/ontology/ 2 http://www.semangit.de/ 3 http://visualdataweb.de/webvowl/#opts=doc=0;editorMode=true;#iri=https://raw. githubusercontent.com/SemanGit/SemanGit/master/Documentation/ontology/semangitontology.ttl

References

1. Comparison of source code hosting facilities . https://en.wikipedia.org/wiki/ Comparison_of_source_code_hosting_facilities, accessed: August 28 , 2019

2. Gitlab gains developers after microsoft buys rival github . https://www.reuters.com/article/us-github -microsoft-gitlab/ gitlab-gains-developers-after-microsoft-buys-rival-github- idUSKCN1J12BR , accessed: August 28 , 2019

3. Auer , S. , Bizer , C. , Kobilarov , G. , Lehmann , J. , Cyganiak , R. , Ives , Z. : Dbpedia: A nucleus for a web of open data . In: Semantic Web , pp. 722 { 735 . Springer ( 2007 )

4. Auer , S. , Lehmann , J. , Hellmann , S. : Linkedgeodata: Adding a spatial dimension to the web of data . In: ISWC . pp. 731 { 746 . Springer ( 2009 )

5. Auer , S. : Towards an open research knowledge graph ( Jan 2018 ), https://doi. org/10.5281/zenodo.1157185

6. De Nies, T. , Magliacane , S. , Verborgh , R. , Coppens , S. , Groth , P.T. , Mannens , E., Van De Walle, R.: Git2prov: Exposing version control system content as w3c prov . In: ISWC (Posters & Demos) . pp. 125 { 128 ( 2013 )

7. Dumbill , E.: Doap: Description of a project . http://trac.usefulinc.com/doap ( 2010 )

8. Kubitza , D.O. , Bockmann, M. , Graux , D.: SemanGit: A linked dataset from git . In: Proceedings of 18th International Semantic Web Conference ( 2019 )

9. Li , J. , Wade , V. , Sah , M. : Developing knowledge models of social media: A case study on linkedin . Open Journal of Semantic Web (OJSW) 1 ( 2 ), 1 { 24 ( 2014 )

10. Torvalds , L. , Hamano , J.: Git: Fast version control system . http://git-scm. com ( 2010 )

11. Vandenbussche , P.Y., Atemezing , G.A. , Poveda-Villalon , M. , Vatant , B. : Linked open vocabularies (lov): a gateway to reusable semantic vocabularies on the web . Semantic Web 8 ( 3 ), 437 { 452 ( 2017 )