project:big_data_analytics

We try to analyse bibliographical data using big data technology (flink, elasticsearch, metafacture).

Here a first sketch of what we're aiming at:

We use bibliographical metadata:

Swissbib bibliographical data https://www.swissbib.ch/

  • Catalog of all the Swiss University Libraries, the Swiss National Library, etc.
  • 960 Libraries / 23 repositories (Bibliotheksverbunde)
  • ca. 30 Mio records
  • MARC21 XML Format
  • → raw data stored in Mongo DB
  • → transformed and clustered data stored in CBS (central library system)

edoc http://edoc.unibas.ch/

  • Institutional Repository der Universität Basel (Dokumentenserver, Open Access Publications)
  • ca. 50'000 records
  • JSON File

crossref https://www.crossref.org/

  • Digital Object Identifier (DOI) Registration Agency
  • ca. 90 Mio records (we only use 30 Mio)
  • JSON scraped from API

Swissbib

Librarian:

- For prioritizing which of our holdings should be digitized most urgently, I want to know which of our holdings are nowhere else to be found.

- We would like to have a list of all the DVDs in swissbib.

- What is special about the holdings of some library/institution? Profile?

Data analyst:

- I want to get to know better my data. And be faster.

→ e.g. I want to know which records don‘t have any entry for ‚year of publication‘. I want to analyze, if these records should be sent through the merging process of CBS. Therefore I also want to know, if these records contain other ‚relevant‘ fields, defined by CBS (e.g. ISBN, etc.). To analyze the results, a visualization tool might be useful.

edoc

Goal: Enrichment. I want to add missing identifiers (e.g. DOIs, ORCID, funder IDs) to the edoc dataset.

→ Match the two datasets by author and title

→ Quality of the matches? (score)

elasticsearch https://www.elastic.co/de/

JAVA based search engine, results exported in JSON

Flink https://flink.apache.org/

open-source stream processing framework

Metafacture https://culturegraph.github.io/, https://github.com/dataramblers/hackathon17/wiki#metafacture

Tool suite for metadata-processing and transformation

Zeppelin https://zeppelin.apache.org/

Visualisation of the results

Usecase 1: Swissbib

Usecase 2: edoc

  • Dominique Blaser
  • Jean-Baptiste Genicot
  • Günter Hipler
  • Jacqueline Martinelli
  • Rémy Meja
  • Andrea Notroff
  • Sebastian Schüpbach
  • T
  • Silvia Witzig
  • project/big_data_analytics.txt
  • Last modified: 2017/09/18 10:12
  • by andrea