Integrating Data for Archaeology Dimitris Gavrilis, Eleni Afiontzi, Johan Fihn, Olof Olsson, Achille Felicetti, Franco Nicollucci, Sebastian Cuy

Introduction Traditional projects in Archaeology focused on aggregating data into one single format / system Provide users with a unified interface Improve search and retrieval Improve retrieval semantics through specialized metadata schemas ARIADNE goes one step further : data integration Try to model the domain information (ARIADNE Catalog Data Model) Use a curation aware aggregator to enrich information using the above model Improve user experience through more substantial and powerful queries

Innovation Why hasn’t anyone done this before ? Complexity Performance Domain knowledge Standard aggregation systems / architectures are insufficient.  ARIADNE Infrastructure

ARIADNE Infrastructure Flexibility Ingest diverse and heterogeneous data XML, RDF, Excel, CSV, … Handle each datastream independently and according to it’s requirements Adapting aggregation, validation, enrichment workflows Add new curation services easily and on demand

ARIADNE Infrastructure Complexity De-couple services complexity through a micro-service oriented architecture Use loosely connecting services in a highly scalable environment. Performance Scalable technologies

ARIADNE Infrastructure Domain knowledge Integrate the domain model (ACDM) into the infrastructure Make extensive use of domain thesauri (e.g. AAT) and label every resource accordingly Create specialized micro-services for curating content according to the domain needs

Data Integration Overall Architecture Repository MORe RDF Store (RDF) Validation Integration Experiments Cleaning RDF Store (CRM) Excel Sheet ARIADNE Portal Enrichment Elastic Search ARIADNE Registry Integration Archive

Use of RDF Every resource is assigned a unique and persistent identifier that is resolved through a URI Every resource has an RDF representation according to the ACDM schema

Data Curation Use of curation micro-services for enriching content Geo-normalization (identify, extract and normalize places and coordinates) Geo-coding (e.g. Geo-names) Thesauri mappings (map native subject terms to a common thesauri : AAT) Temporal normalization (identify, extract and normalize dates) Gazetteers (e.g. DAI Gazetteer) Historical & Ancient place names identification (Pelagios & Pleiades) Temporal information mappings (Perio.do)

Data Integration Data Integration is based on a 3+1 dimensions Subject Space Time Resource type

Identify & Link together Resource Types Model individual information resource types (e.g. collections, bibliographic reports, databases, datasets, etc). Identify each resources type during ingestion Link / group different resource types E.g. put all related heterogeneous resource types (reports, datasets,…) under the same collections

Thematic integration ARIADNE uses the AAT thesaurus to semantically label ALL aggregated information. AAT terms act as a glue and when combined with spatial and temporal information can produce great results Semantic expansion of terms is extensively being used in order to improve retrieval. Expansion of multi-lingual terms facilitates cross-language search without requiring automatic translation.

Spatial & Temporal All resources with spatial information Are assigned WGS84 projected coordinates All resources with temporal information Are normalized according the ACDM dates (that takes into account periods, period names and supports ISO date format).

Subject Terms Curation Lifecycle *nativeSubjects AAT Native Subjects mappings Vocabulary Mapping Tool MORe Provider Native Repository *nativeSubjects *providedSubjects Excel Sheet XML Files Registry *nativeSubjects *providedSubjects ACDM / Subjects (JSON) **providedSubjects **derivedSubjects **broaderGenericSubjects *nativeSubjects ARIADNE Portal Elastic Search *mono-lingual (prefLabel only) ** multi-lingual (prefLabel & altLabel)