Semantic Scholar

Semantic Scholar provides:

• academic searching across its collection of research papers
• filters for narrowing search results
• links to papers, authors, references and citing publications
• suggestions for related papers
• author pages and publication histories
• citation counts and automated citation indicators
• AI-generated summaries and answers for some papers
• saved libraries, research feeds and alerts for registered users

The service describes itself as a tool for scientific literature. Coverage varies between subjects and types of publication, so it may be more useful in some areas than others. It is not a dependable route for finding academic books, book chapters, legal sources or other specialist materials.

Semantic Scholar may help when you want to:

• try a different route to papers on a topic
• locate a particular paper or author
• explore papers related to one you have already found
• follow references to earlier research
• find later papers that have cited a publication
• scan a collection of papers before deciding which ones to read
• receive recommendations based on papers you have saved
• explore scientific or technical literature for an extended project

It can complement other search tools, but you do not need to add it automatically to every search. The tools you use should reflect your assessment brief, subject, level of study and research method.

You can search Semantic Scholar using a topic, paper title or author. Quotation marks can be used to search for a phrase.

Semantic Scholar does not support Boolean operators or wildcards. If you need to construct a precise search using AND, OR, truncation, controlled vocabulary or particular search fields, use LibrarySearch or an appropriate specialist database.

Use the filters in Semantic Scholar to narrow results by criteria such as date, subject area or publication type.

When you find a useful paper, its record may provide:

• References, showing earlier work cited by the paper
• Citations, showing later papers that have cited it
• suggestions for related papers
• links to author pages
• links to available versions or publisher pages

Following references backwards and citations forwards is known as citation chaining. See Citation chaining for guidance on using this technique.

AI-supported features are not available for every paper and may change as the service develops.

TLDR summaries

A TLDR is a short, automatically generated summary of a paper’s main objective and findings. It can help you decide whether a paper may be worth investigating.

Do not use the TLDR as a substitute for the abstract or full paper. A one-sentence summary can omit the research context, methods, limitations, uncertainty or qualifications that affect what the findings mean.

Ask This Paper

Ask This Paper lets you ask questions about an individual paper and receive an AI-generated answer with supporting statements drawn from the paper.

Supporting statements give you material to check, but they do not prove that the answer is accurate. Read the relevant part of the original paper and check that the answer has represented it correctly and in context.

Semantic Reader

Semantic Reader is an enhanced reading view available for some papers. It can provide contextual information and help you move between the paper and connected research.

Any generated explanation or additional information should still be checked against the original paper and appropriate academic sources.

Using Semantic Scholar to locate papers is different from using its generative AI features to create summaries or answers. If you use these features as part of assessed work, follow your assessment brief and acknowledge the use of AI where required.

For fuller guidance, see:

• AI for researching a topic
• Evaluating AI outputs

Semantic Scholar shows references, citing papers and citation counts. It may also identify some citations as highly influential using automated analysis.

These indicators can help you decide what to investigate, but they are not measures of academic quality:

• a highly cited paper is not necessarily reliable or relevant
• a recent paper may be important but have few citations
• being cited does not mean that later researchers agree with a paper
• an automated influential-citation label may be incomplete or incorrect
• citation patterns vary considerably between subjects

Open the citing paper if you need to understand how it has used the earlier work. Check whether it supports, challenges, applies or simply mentions the source.

Semantic Scholar may link to an openly available PDF, a repository copy or a publisher’s page. However, finding a record does not mean that the full text is available through Semantic Scholar.

Semantic Scholar does not connect directly to Abertay’s subscriptions. If you find a paper that you want to read:

1. Search for its title in LibrarySearch.
2. Check whether Abertay provides access or whether an open-access version is available.
3. Try LibKey Nomad if you are viewing the paper on a publisher or other external website.
4. Follow the guidance on accessing full-text journal articles if you still cannot open it.

The absence of full text in Semantic Scholar does not mean that Abertay cannot provide access.

Different tools may find overlapping material, but they organise and present it differently.

• LibrarySearch searches Abertay’s books and e-books alongside a broad collection of articles, conference papers, theses and open-access research. It connects results to Abertay access and provides academic search filters.
• Google Scholar searches scholarly material across publishers, repositories, universities and other websites. It is particularly useful for broad discovery, finding alternative versions and following citations.
• Semantic Scholar uses AI-supported features to organise research papers, recommend related work and provide generated summaries or answers for some papers.
• Generative AI research assistants are generally more question-led and may create multi-paper answers, comparisons, reports or extraction tables.

These boundaries are becoming less distinct as services add new features. Choose a tool for what it helps you do rather than assuming that one search service covers everything.

Semantic Scholar is useful for discovery, but:

• its coverage is not comprehensive and varies between subjects
• books, chapters and some specialist or professional sources may be poorly represented
• it does not support Boolean operators or wildcards
• records, citation counts and author information can contain errors
• preprints and other material that has not been peer reviewed may appear in the results
• automated recommendations depend on the papers and activity used to generate them
• AI-generated summaries and answers can omit context or misrepresent the original
• citation and influence indicators do not establish quality or reliability
• the service does not provide a fully controlled or reproducible search process

Do not use Semantic Scholar alone when your research method requires a comprehensive, systematic or reproducible search. Follow the databases and search methods required by your discipline, supervisor or review protocol.

See Evaluate information for guidance on assessing the original sources you find.