I've been doing some work with colleagues at Cambridge and Imperial over the last year on using LLMs to improve evidence synthesis, primarily trying to find papers on the effectiveness of certain Conservation interventions. It's becoming clear that you really need to move beyond screening papers only by title and abstract - there's often information buried deep within papers that can only be found with access to full text. My colleague Anil Madhavapeddy has written a bit about our adventures in trying to ingest full-text academic papers: https://anil.recoil.org/notes/uk-national-data-lib
Yes, it depends on what you're doing; for general paper discovery / search tasks, title abstract can be enough (which is also why Springer and Elsevier have been pulling even their abstracts from sources like OpenAlex).
But for something like that you need full texts to look into results sections. I'm very curious how you're dealing with information contained in tables, or if you're dealing with snippets of text from the full-text alone. Have you poked around Elicit yet?
I've recently had this problem where the important information (number of study participants, and how many were filtered out during which step) were only encoded in figures, not in the text. Maddening.
I've been doing some work with colleagues at Cambridge and Imperial over the last year on using LLMs to improve evidence synthesis, primarily trying to find papers on the effectiveness of certain Conservation interventions. It's becoming clear that you really need to move beyond screening papers only by title and abstract - there's often information buried deep within papers that can only be found with access to full text. My colleague Anil Madhavapeddy has written a bit about our adventures in trying to ingest full-text academic papers: https://anil.recoil.org/notes/uk-national-data-lib