- Collaborators: Dani Fernández, Dani Gayo, José Labra.
MERA (Musical Entities Reconciliation Architecture) consist of an architecture highly configurable designed to link music-related datasuources using string comparison techniques and semantic technologies. wMERA is a prototype that implements most of MERA's features, including:
- Graph navigation exploring alternative forms for each entity.
- Graph navigation exploring related entities.
- Configuration of relevancies when applying refinements to a query.
- Configuration of minimum acceptable values for each type when scoring a result.
- Blocking function: adaptation of q-gram indexing and TF-IDF.
- Usage of MERA rdf graph schema
- Set of comparison algorithms of general-purpose.
- Set of text standardization functions.
wMERA has been developed in python 2.7 and requires the next packages:
- rdflib
- pymongo
The code is organized as follows:
- Package wmera: Main package. It contains the code of the entity reconciliatory and the modules involved in that process.
- Sub-package adapters: conversion between different implementations of q-gram indexes.
- Sub-package controller: code to coordinate query execution with result retrieving.
- Sub-package graph_gen: code to generate RDF graphs through consuming an interface of a parser that yields model objects. It also contains an implementation of MERA’s graph interface using RDFLib library.
- Sub-package infrastructure: interface and implementations of the different q-gram indexes.
- Sub-package parsers: interface of parsers that generate model objects as well as implementations to generate objects of different datasources.
- Sub-package query_gen: code to generate serialized MERA queries in a JSON model and after parsing another JSON model.
- Sub-package mera_core: model objects, system interfaces, matching module and string comparison packages.
- Module facade: code to consume MERA from external apps.
- Module factory: code to build some specific complex objects.
- Module utils. Different utility methods.
- Module word_utild: different utility methods focused in strings.
- Package apps: scripts that use somehow wmera code: research experiments, graph generation, comparison of tehcniques…
- Package test: testing code for wmera. Most of these test are integration test, and they expect to be executed in a machine running an instance of MongoDB accessible at localhost:27017.
Documentation of MERA and wMERA are still in progress. Contact with Dani Fernández for any question.