Building a thesis repository with embargoes, restricted material and a publisher objection
A state university has 38 affiliated colleges and has produced about 1,100 doctoral theses since 1995. In compliance with the UGC's minimum standards and procedure for PhD, the university has signed the Shodhganga memorandum with INFLIBNET and must deposit electronic versions of all theses. The library has inherited an Excel list of thesis titles, an unknown number of PDFs on staff laptops, and 240 scanned theses, of which 60 are image-only scans without OCR. Fourteen theses are under an injunction from the university's research committee because patents are pending; nine contain Survey of India maps marked 'restricted'; one has been published as a book by a commercial publisher that has written objecting to open access. Researchers want to be able to restrict access to their own theses for two years. The university also wants the repository to be harvestable by the National Digital Library of India and to appear in Google Scholar.
1. Which metadata elements and identifier scheme would you require, and why?
Read the worked answer
Mandatory elements (qualified Dublin Core, which is what DSpace stores natively): title; author; advisor and department as contributors; date issued; type ('Doctoral thesis'); publisher (the university); language; subject keywords plus the class number; abstract; rights; and the persistent identifier.
Interoperability layer: expose OAI-PMH with both oai_dc for generic harvesters such as the National Digital Library of India and the ETD-specific schema etdms, which carries degree, degree level, advisor and department in fields that generic harvesters lose. Define OAI sets by department and by award year so harvesting can be selective.
Identifiers: mint a handle at accession, since DSpace does this by default and it is stable. A DOI through DataCite only if the university will commit to maintaining the resolver and paying the fee indefinitely - a dead DOI is worse than no DOI.
Google Scholar indexing depends on crawlable, server-rendered item pages with citation_title, citation_author, citation_publication_date and citation_pdf_url meta tags, plus a sitemap. A heavily JavaScript-dependent interface will be poorly indexed, so test by fetching a page as a crawler would before promising visibility.
2. How would you handle the restricted maps, the patent cases and the publisher's objection?
Read the worked answer
Patent cases (14): metadata-only records. The bibliographic record is public and citable; the file is withheld under a dated research-committee decision with a review date, and held offline in a controlled store with a checksum. The committee, not the library, owns that decision - get it in writing.
Restricted Survey of India maps (9): these carry their own access conditions independent of copyright. Do not publish the full file. Either obtain clearance or deposit a redacted version with the restricted sheets removed, and record what was removed and why.
Publisher objection (1): read the publication agreement before responding. A thesis is normally a prior work and the author usually retains the right to deposit the version that was examined; the publisher's exclusive licence typically covers the published version, not the thesis. Deposit the examined thesis, not the publisher's PDF or the published book text, and record which document was consulted. If the agreement is genuinely exclusive and retroactive, negotiate or apply an embargo rather than capitulate by default.
Two-year embargoes requested by researchers: supported natively. Keep metadata visible so the work remains citable, embargo the bitstream with an end date, and run an automated release job - manual release will be forgotten. Record the requester, the reason and the end date.
3. What workflow would you set for the 60 image-only scans and the inherited Excel list?
Read the worked answer
Do not build on the Excel list. Reconcile against the registrar's award register and the degree records - those are authoritative for title, author, year, advisor and department. The spreadsheet is a finding aid, not a source of truth, and any thesis it omits will otherwise vanish from the repository.
Scans: OCR the 60 image-only files, then quality-check a sample for accuracy; produce a searchable viewing PDF alongside a PDF/A preservation master. Note in the record that OCR was applied, since OCR errors affect search results.
Batch load: import by CSV with a controlled vocabulary for department and degree, deduplicate on title + year + author (allow fuzzy matching, the same thesis is often listed twice under variant titles), and check advisors for missing values.
Quality assurance: verify a 10% random sample against the physical record before the collection is published; mint handles; confirm that every deposited thesis required by the memorandum is actually present and that the count matches the registrar's list.
Publication: go live in stages - completed metadata first, then open files, then embargoed items - so that errors are found while the collection is small. Publish the deposit workflow to departments so future theses arrive with correct metadata instead of being reconstructed later.