Methodology

How records get into MySCI, what is measured, and what is still missing. Published so that a reader can check what "indexed" means here rather than taking the word on trust.

How a journal becomes indexed

1. Registered. The journal is recorded with its publisher, identifiers, and web presence.
2. Endpoint verified. MySCI contacts the journal's OAI-PMH endpoint, confirms it is a working repository, and records which metadata formats it offers.
3. Harvested. Article records are retrieved page by page. Each response is retained verbatim before anything is parsed from it.
4. Ingested. Records become articles, authorships, and keywords. Re-harvesting is safe: identity is keyed on the repository's own identifier and change is detected by checksum, so nothing duplicates.
5. Monitored. Later harvests pick up corrections. Earlier versions of a record are kept rather than overwritten.

Where the registry stands

What limits coverage today

MySCI indexes 0 of 0 known journals. The gap is not an oversight; it is what has actually been harvested so far. Beyond that, four faults in real repositories limit what can be collected, and each is recorded against the journal rather than hidden:

• Truncated record lists. Some repositories signal the end of their records while their own counters say more remain. MySCI stops — continuing would mean fabricating a request the protocol says has no interpretable structure — and reports the shortfall.
• Unstable pagination. Some repositories return a different set of records on each pass, so one harvest cannot see everything. Repeated harvests converge, and the journal is marked partially indexed until they do.
• Malformed metadata. A single character that XML forbids can make an entire page of records unreadable. MySCI removes only illegal characters, records that it did so, and marks every affected record.
• Contradictory records. A repository may report the same article as both published and deleted. MySCI keeps the published version and records the contradiction.

Onboarding: how a journal gets an endpoint

A journal can only be harvested once MySCI knows its OAI-PMH address. Those addresses are derived from each journal's homepage and then proved before use — a derived URL is a guess, and a guess is only acceptable because it has to answer for itself first.

One guess is refused even when it would work. In OJS, /index.php/index/oai serves every journal on an installation, and dozens of journals here share a server. Accepting it for one journal would harvest the whole site into that journal — successfully, with correct provenance, and undetectably. So a URL addressing a whole installation is refused unless the repository proves it holds a single journal.

Counted per candidate URL, not per journal: most journals take more than one attempt, and the refusals are the part that explains why a journal has no endpoint.

Endpoint status across the registry

Metadata coverage measures

Journal pages report field coverage as separate components — DOI, abstract, keywords, licence, and so on. There is deliberately no single quality score. A composite would need a published formula, tested weights, and a check for disciplinary bias before it could mean anything, and none of that exists yet.

Coverage is only reported for fields the harvested metadata format can actually express. Unqualified Dublin Core, which nearly every Malaysian OJS journal exposes, has no element for an author identifier, an affiliation, or a reference list. Reporting those as 0% would criticise journals for a limitation of the format, so they are shown as not measurable instead.

Recognition data

MyCite recognition is imported from a published compilation of the MyCite list. Each record stores its source, the date it was checked, and the method — in this case a third-party compilation rather than a direct reading of MyCite's own system, which is a weaker form of evidence and is labelled as such on every record.

Journals are matched to the list by ISSN and never by title alone. Where the source's ISSNs had been damaged by passing through a spreadsheet, they were repaired only when the corrected value produced a valid check digit.

MySCI also holds 0 third-party indicator value(s) — CiteScore, SJR, Journal Impact Factor, and similar. These belong to Elsevier, Scimago, and Clarivate, and are held for internal analysis with attribution. They are not shown on public pages and are not redistributed.

Citations

Reference lists are not yet supplied by any publisher in the index. Where they exist here they were obtained from OpenAlex, and every reference and every citation link records which source supplied the evidence, so publisher-deposited references can be told apart from third-party ones once journals begin depositing them.

Citations counted within the MySCI corpus and citations reported by external services measure different things over different corpora. They are never added together, and no combined total is stored anywhere. Citation relationships between indexed articles are shown as links a reader can follow and check; no citation count is published, in either universe.

API

Everything on this site is available as JSON at /api/v1/, with reference documentation generated from the code rather than written alongside it, so it cannot drift.

The API is read-only and open, and carries the same qualifications this page does: every article states where its metadata came from, every DOI states whether it is actually registered, an author name is not treated as an identity, and MySCI indexing is never merged with recognition by any external body.

Author identity

Authorships record the author name and affiliation exactly as the publisher supplied them. An authorship is linked to a researcher identity only when the metadata carries a persistent identifier with a valid checksum. Names alone are never sufficient: patronymic constructions, inconsistent romanisation, and common name stems make name matching unreliable in this corpus, and a wrong merge attributes one researcher's work to another.