About this Keyword Finder
This tool helps you find relevant keywords or 'descriptors' from the ERIC Thesaurus based on the content of your uploaded report. The thesaurus contains over 12,000 terms, including 'preferred' and 'non-preferred' terms, and synonyms. The ERIC thesaurus is regularly revised, but was developed for users in the USA, and this is reflected in the terminology (and spelling) that is used.
The tool uses a large language model (LLM) to suggest keywords based on the content of your report. The default LLM provider is OpenAI. If you prefer, you can select DeepSeek, a Chinese Open-Source LLM. Your document is not stored or used for training purposes by either provider and there you do not have to provide any personal information to use this service.
There are three stages in the process:
- First, your selected LLM is used to suggest possible keywords based on the content of the uploaded report (this needs to be a .txt, .docx, or .pdf file). The LLM's both include the thesaurus, but they may not always suggest the 'preferred' terms. Then these suggestions are checked against the thesaurus itself, and 'preferred' terms may be suggested. If the suggested term is not in the thesaurus at all, a list of possible terms which are semantically similar to the suggested term are offered. These may be relevant; but they are just suggestions.
- Second, you can review the LLM's suggestions, consider alternatives, and decide to 'Keep' each suggested term (even if it is not in the ERIC Thesaurus); 'Swap' it for the preferred or suggested term; or 'Drop' it altogether.
- Finally, can copy or download the resulting list of keywords, which you can edit or format as you wish and add to your report or metadata record.