Universidad Americana (Paraguay): cloud-hosted DSpace institutional repository with AI-assisted metadata

Institutional context

Universidad Americana is a private university based in Asunción, Paraguay. Like many institutions in the region, it came to the project with accumulated academic output (theses, final projects, articles and institutional documents) that existed as files rather than as a published, searchable collection.

The challenge

The goal was twofold, and both parts mattered equally. The first was to put online an institutional repository that would bring together the university's academic output under international standards, with persistent identifiers and with metadata that academic search engines and aggregators could harvest. The second was that this repository should not depend on infrastructure the university would have to provision, secure and maintain on its own.

There was also a well-known risk in this kind of project: a new repository is born with an initial loading debt. If describing each accumulated document requires manual transcription, the repository opens with fifty items and stays there. That is the pattern behind so many repositories in the region that are formally active and practically empty.

What we implemented

• DSpace as the open-access institutional repository, hosted in the cloud on infrastructure managed by BiblioLabTech.
• MetaExtract API integrated into the deposit workflow, to auto-fill the descriptive metadata of every incoming document.

How we did it

1. Defining the information model. The structure of communities and collections was agreed according to the university's academic organization, along with the metadata schema and its mandatory fields, the document types and the usage licenses.
2. Cloud deployment. The infrastructure was provisioned with HTTPS, automatic backups and monitoring, with no servers or operations staff required on the university's side.
3. Initial content loading. The accumulated output was deposited with its files and metadata, with MetaExtract API speeding up description and human review before each item was published.
4. Customization and interoperability. The interface was adapted to the university's identity, and OAI-PMH exposure, metadata for academic search engines and harvesting by regional aggregators were configured.
5. Training and launch. The library team was trained in collection administration, metadata review and approval workflows, and the repository was opened to the public.

Result

The repository is online and publicly verifiable: it appears among the active implementations that BiblioLabTech lists on its website. The DSpace contract with MetaExtract API was renewed for a new period, and the university is evaluating an expansion toward DSpace-CRIS, which would add research information management on top of the existing repository. That progression, from repository to research information system, is the natural path when an institution starts needing to report its output by researcher and by unit.

Technologies

DSpace, MetaExtract API, Dublin Core, OAI-PMH, Handle, cloud hosting managed by BiblioLabTech, and DSpace-CRIS as an expansion under evaluation.