When I annotate a scene, how do I record who asserts or believes each statement and how committed each party is, so that the narrator's statement and a character's belief are not collapsed into one 'fact'?

FactBank 1.0's guidelines give the most explicit open-text model of source-relative commitment: every event has a default source (the text author, s0), reported or believed content adds a nested source such as 'John according to the author', and each source gets its own value (certain/probable/possible, positive/negative/underspecified). The guidelines' own examples show the author uncommitted while a reported source is certain, and the reverse (author asserts, character ignorant). The scheme is built on newswire, records no separate 'analyst inference' layer (footnote 2 excludes the annotator's own beliefs), and was not applied to fiction by this source; mapping the narrator onto s0 is the analyst's step (synthesis). Curator qualifications: FactBank source-relative factuality: author s0 default source; nested sources (John according to author) with separate commitment values; annotator belief excluded by design. 8/8 excerpts located. Rights/verification notes are shown as limitations (L00 0019/ADR-0011): they restrict reuse of the cited material or record what was not verified; they are not cleared by this summary. Limitations: [component_reuse_restriction] FactBank corpus is LDC-licensed; only the guidelines are open. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.

{
  "citations": [
    {
      "url": "https://www.cs.brandeis.edu/~roser/pubs/fb_annotGuidelines.pdf",
      "note": "candidate evidence excerpts"
    },
    {
      "url": "https://catalog.ldc.upenn.edu/LDC2009T23",
      "note": "candidate evidence excerpts"
    }
  ]
}