{
  "annotation_id": "d6e3e70a-b08a-43b9-ae80-d56ff7138888",
  "date_created": "2026-03-02T18:01:04.040000Z",
  "date_modified": "2026-03-02T18:01:04.040000Z",
  "file_hash": "22b2382aa4afa793dd9019794e23c4f5b063db2a8ac972ce66f8690ce0449c30",
  "private": false,
  "record": {
    "abstract": "Understanding texts requires memory: the reader has to keep in mind enough\nwords to create meaning. This calls for a relation between the memory of the\nreader and the structure of the text. To investigate this interaction, we first\nidentify a connectivity matrix defined by co-occurrence of words in the text. A\nvector space of words characterizing the text is spanned by the principal\ndirections of this matrix. It is useful to think of these weighted combinations\nof words as representing ``concepts\u0027\u0027. As the reader follows the text, the set\nof words in her window of attention follows a dynamical motion among these\nconcepts. We observe long range power law correlations in this trajectory. By\nexplicitly constructing surrogate hierarchical texts, we demonstrate that the\npower law originates from structural organization of texts into subunits such\nas chapters and paragraphs.",
    "arxiv_id": "physics/0510276",
    "authors": [
      "E. Alvarez-Lacalle",
      "B. Dorow",
      "J. -P. Eckmann",
      "E. Moses"
    ],
    "categories": [
      "physics.soc-ph",
      "cond-mat.stat-mech"
    ],
    "title": "A quantitative analysis of concepts and semantic structure in written   language: Long range correlations in dynamics of texts",
    "url": "https://arxiv.org/abs/physics/0510276"
  },
  "schema_id": "dorsal/arxiv",
  "source": {
    "execution_id": "31bad076-f97a-46e4-80ee-f9dfbec2df72",
    "id": "arXiv Dataset IDs",
    "type": "Model",
    "variant": "snapshot-2026-03-01",
    "version": "0.1.0"
  },
  "user_id": 1000002
}