Citation and patent databases record which research papers cite which other papers, and list the world's patents, so that people can judge research and see who is inventing what. A citation is a reference in one paper to an earlier paper that it builds on. A database of citations is called a citation index.
How is a citation index built? The provider chooses a list of journals that it considers reliable. For every new paper in those journals, it records the authors, their institutions, the journal and the date. Then it reads the paper's list of references, and links each reference to the paper it points to. This step is hard. References are written in many styles, with spelling mistakes and short forms, and different authors with the same name must be told apart. Once the links exist, the database can count how many times the work of any paper, author, university or country has been cited.
These counts are used in many decisions. Universities use them when they hire and promote faculty. University rankings use them. Funding bodies use them when they choose projects. A journal's impact factor is the average number of citations that its recent papers received, and authors use it to choose where to send a paper. But a citation count measures attention, not quality, and some groups cite each other only to raise their numbers. So providers check for unusual citation patterns. Legal research has a similar tool for judgments, called a citator.
Say the research office of a private university in Tiruchirappalli is preparing its figures for a national ranking. It searches the citation index for papers from the last five years with at least one of its faculty as an author. The search returns 2,340 papers, cited 18,700 times in all. But 60 papers are missing, because their authors spelt the university's name in three different ways. The office asks the provider to merge the three spellings into one institution record, and the count is corrected to 2,400 papers.
Patent databases do the same work for inventions. Each patent cites earlier patents and papers. A company searches the database before it files its own patent, to check that its idea is new. It also searches to see which rivals are filing patents in its field. For instance, a battery maker can see that a rival filed 40 patents on cell cooling last year.
Both kinds of data come from the world of academic publishing. Journal publishing platforms send the provider the details of each new paper. Financial research platforms also search patents for investors who study technology companies.