Linking IASSIST 2017 Presentations With Research Papers

The archived IASSIST 2017 site preserves the “Data in the Middle: The Common Language of Research” conference, held at the University of Kansas in Lawrence from 23–26 May 2017. Its programme, plenaries, check-in guidance, transport notes, local recommendations and presentations on big data, deep learning, digital agriculture and global food security remain valuable. Yet readers often need to move between a session page, a speaker name and a separate paper repository to reconstruct the research trail. The archive is informative for people, but less effective for machines comparing records across events and publication venues.

Linked Data offers a practical way to make that trail explicit. Stable web identifiers and RDF triples can relate a presentation to its session, speaker, topic, slides and eventual journal article. The resulting linked open data layer supports richer search, cross-referencing and reuse without changing the familiar pages in the archive. It is especially relevant to Australian scholars, who often work across local collections, national research services and international publishing systems.

Why Archival Programmes Need Connected Research

A conference programme works well as a chronological guide, but its structure does not always express the relationships that make scholarship useful. A talk may extend a paper, repeat a pilot reported at another meeting, or share data with work published months later. Readers following digital agriculture or deep learning should be able to move from a talk to its evidence without guessing which names and titles refer to the same work.

The archived IASSIST programme supplies sessions, times, rooms and presenters, while presentation pages carry titles and descriptive details. An ordinary hyperlink can connect two pages when an editor knows the destination, but it cannot consistently state whether two items are versions, related outputs or separate contributions. It also asks visitors to follow a trail designed for navigation rather than automated retrieval.

A linked-data layer can retain ordinary hyperlinks while adding typed relationships and identifiers that software can query. This semantic web approach supports discovery by topic, organisation and shared contributor. It also makes it possible for a new scholarly graph to draw on the archive without copying every page or changing its presentation. The goal is interoperability: clear links between people, events and outputs.

A Linked Data Model for Scholarly Records

At the centre of the model should be persistent URIs for the conference, its sessions, each contribution and the people involved. RDF expresses these links as subject–predicate–object statements: a presentation has a creator, belongs to a session and addresses a subject. A paper can then be linked to the event that contextualised it. The conference, its archived website and an individual page remain separate entities because each has a different identity and purpose.

Reusing a shared vocabulary makes the graph legible to other systems. Schema.org can describe events, people and scholarly articles, while Dublin Core supplies basic descriptive properties. A small event ontology can add classes for sessions, presentations and contributions, connecting them to established standards. The sameAs predicate is useful when an external identifier denotes the same entity, but it should not collapse a conference talk and the paper it discusses.

Persistent identifiers provide dependable anchors for these connections. DOIs can identify papers, ORCID identifiers distinguish researchers, and Research Organization Registry identifiers organise institutions. Internal URIs remain necessary for conference sessions and presentations that lack external identifiers. Provenance should accompany every mapping, recording the source page, responsible curator, creation date, review date and any uncertainty in the match.

Linking Event Entities Across the Archive

Begin with one canonical record for the event and assign internal identifiers to every session and contribution. Each archived page should point to that entity, while the entity points back to the page that serves human readers. This creates a dependable graph and prevents a renamed title or copied PDF from becoming another unconnected object. Original URLs remain useful entry points rather than the only identity.

Relationships should carry specific meanings that software can act upon. Predicates such as presentedBy, partOf, hasTopic and isAssociatedWith can connect a person, a session, a subject and a research output. A slide deck can be described as basedOn a paper, while a later article can be connected to the earlier presentation through an associated-work relationship. Consistent predicates make these links more useful than a collection of generic “related items” links.

Not every related object is identical. A conference presentation, a preprint, a proceedings paper and a later journal article should remain separate entities even when their titles or arguments are closely related. The model can express version history, shared authorship and derivation without claiming that the records are interchangeable. This distinction protects scholarly accuracy while still giving readers a navigable path through the research.

Comparing Metadata Patterns and Query Value

The practical differences between conventional web navigation and a linked-data publishing approach can be summarised directly:

Aspect	Basic web linking	Linked Data publishing
Identity	A page URL acts as the label	Persistent URIs and explicit equivalence
Relationships	Readers see a generic “see also” link	Typed, machine-readable predicates
Discovery	Manual browsing and keyword search	SPARQL queries across connected datasets
Reuse	Links are copied into another system	Structured export and federated lookup
Quality	Confidence is inferred from wording	Provenance, dates and confidence are recorded

This approach moves editorial effort from repeated manual link checking to reusable metadata. A curator can correct a speaker URI once, and every session, slide deck and paper connection using it can reflect that correction. Plain HTML remains available for people and aggregators that do not query RDF, while the graph supplies precision for systems that do.

Shared vocabularies also make the archive easier to combine with bibliometric datasets. Presentation topics can align with controlled subject terms, organisations through ROR identifiers, and papers through DOI records. The semantic layer becomes a bridge between the conference schedule, research outputs and external indexes rather than another isolated catalogue.

Discovering Papers Through Shared Identifiers

Connecting a presentation to a paper requires more than matching title strings. Strong signals include author identifiers, normalised titles, DOI metadata, conference names, dates and shared subject terms. A title match alone can link unrelated work, while a DOI or consistent author identifier provides stronger evidence. Mapping software should combine these signals, record its reasoning and allow a curator to approve uncertain relationships.

A related event source such as the AIC 2017 programme can provide a useful comparison when designing the same event model. Rather than copying one conference’s page structure, publishers can express both sets of records with shared classes for sessions, talks, people and organisations. This creates options for federated search while preserving each programme’s original context and editorial judgements.

With those links in place, a SPARQL query can ask coherent questions across the knowledge graph. It could list presentations connected to papers with DataCite identifiers, find outputs concerning digital agriculture, or show which speakers from the University of Kansas also published on food security. Where matching is uncertain, the response should expose the supporting evidence and confidence level rather than presenting a speculative association as certain.

Designing for Australian Research Contexts

For Australian readers, the semantic layer should fit national discovery and research infrastructure. Records can carry identifiers used by Trove, align with metadata indexed by the Australian Research Data Commons, and use Australian subject terms where available. International vocabularies provide interoperability, while local authority data helps distinguish organisations, projects and datasets with similar names.

Practical information also needs human-readable anchors. In Sydney and Melbourne, conference travel often depends on Opal or Myki, accurate interchange information and accessible venue directions. Stable links to transport and check-in guidance help attendees retrieve practical details long after a conference schedule has circulated through email or social media.

Digital agriculture has a strong local context in Brisbane, where horticulture, family-owned growers and fresh produce markets connect research with everyday food systems. Adelaide’s Central Market and nearby farmers’ markets offer another example of the supply chains examined in agriculture presentations. Controlled concepts for food security, supply chains and farm data can help local datasets connect with relevant conference outputs.

Distance and border conditions matter elsewhere. Perth-based participants may depend on remote access or long travel connections, making stable links to recordings and digital papers especially valuable. Australian Customs and Border Force biosecurity rules may also apply to seeds, plant material or soil brought into the country. Conference guidance should link official requirements and clearly distinguish physical specimens from the datasets that describe them.

A Practical Publishing Workflow

The workflow can be implemented without rebuilding the archive. A small editorial team can publish versioned metadata beside the existing HTML, create URIs for event entities, connect records to external identifiers and expose the result through a public SPARQL endpoint or downloadable graph. Each release should retain a change log and a mapping from former labels to current entities.

• Create stable URIs for the conference, sessions, contributions, speakers and organisations, and retain them when titles or rooms change.
• Use schema.org, Dublin Core and a small event vocabulary; reserve sameAs for genuine identity matches.
• Connect people through ORCID, organisations through ROR and papers through DOI, Crossref or DataCite records.
• Link talks, slides, proceedings and papers with typed relations, preserving distinctions among preprints, versions and later articles.
• Record each mapping’s source page, curator, review date, matching method and confidence, with corrections visible in a change log.
• Publish machine-readable exports and test SPARQL queries against known presentation-to-paper links while retaining accessible human pages.

Together, these practices preserve the programme’s human-readable pages while adding a durable knowledge graph. The result is an archive that remains easy to browse, yet lets presentations, speakers and research outputs be discovered, compared and reused with less dependence on a visitor’s memory or a single search phrase.