provit 1.1.1

A light, dezentralized provenance tracking framework using the W3C PROV-O vocabulary

provit is a data provenance annotation and documentation tool. It provides various feature for creation and retrieval of provenance information for data stored in files. The tracking of sources, modifications and merges allows the user to keep a log of all modifications a dataset was subject to. This is especially useful for dataset which are accessed intermittently or part of a long running workflow (e.g. for a scientific thesis). Furthermore, provenance data stored next to the data in an archive can help others to identify quality, value and acutality of the data.

provit does not require any external infrastructure. All information is stored in .prov files right next to the data files as a JSON-LD graph. This makes it the perfect tool for small teams or individual researchers.

To allow interoperatibility, a small subset of the W3C PROV-O vocabulary is implemented. Therefore, the provenance information can easily be merge in a linked data graph if necessary, at a later stage of the project.

provit aims to provided an easy to use interface for users who have never worked with provenance tracking before. You can operate the tool using the

If you feel limited by PROVIT you should have a look at more extensive implementations, e.g.: prov.

Full documentation is available under: provit.readthedocs.io.

Quick Installation

Note

provit requires a working installation of Python 3.7, furthermore the use of a virtualenv is strongly encouraged. If you need help to set this up, please follow the Installation section in the documentation.

provit is availabe via the Python Package Index (PyPI) and can be installed by using pip pip. Simply create a virtualenvironment with your preferred method a run the pip install command:

$ mkvirtualenv provit
$ pip install provit

Quickstart

provit provides three modes of interaction:

• command line interface

• graphical user interface

• python package

All of them allow you to track provenance, but the provit browser additionally lets you explore tracked provenance.

provit browser

You can start provit browser directly from your terminal:

$ provit browser

provit cli

Simply cd to the directory, where your data is located, create (or append to an already existing) provenance file.

$ provit add FILEPATH [OPTIONS]

The –help command shows you the full list of available options and arguments.

$ provit --help

provit package

Using provit in your ETL pipeline is easy. simply import the Proveance class and start using it (e.g. as displayed below).

from provit import Provenance

# load prov data for a file, or create new prov for file
prov = Provenance(<filepath>)

# add provenance metadata
prov.add(agents=[ "agent" ], activity="activity", description="...")
prov.add_primary_source("primary_source")
prov.add_sources([ "filepath1", "filepath2" ])

# return provenance as json tree
prov_dict = prov.tree()

# save provenance metadata into "<filename>.prov" file
prov.save()

Roadmap

We have a small roadmap, which we will make transparent below:

• Increase test coverage (currently 81%)

• Windows support (all devs are on Linux)

• Agent management in PROVIT Browser

Overview

Authors:

P. Mühleder [e-mail], F. Rämisch [e-mail]

License:

MIT

Copyright:

2018-2019, Peter Mühleder and Universitätsbibliothek Leipzig