Skip to main content

Table 1 Indexes of failure of de iure and de facto disclosures—methodology and ranking

From: Algorithmic disclosure rules

  Relevant failure index Proxy Methodology and Ranking (O/S–O)
Readability Information overload Length of text No. of polysyllables on the basis of the length of the text
Rank: e.g. if longer that X words (golden standard), then rank S–O
ALGO: SMOG; Dale–Chall readability formula; Gunning Fog Index
Major Ref. Bartlett et al. (2019)
Informativeness Information overload Complexity of text Syntactic: No. of certain grammatical structures (nodes) containing complex text (e.g. conjunctive adverbs—however, thus, nevertheless—passives, modal verbs—could, should, might)
Rank: E.g. if number of nodes containing complex tokens in clause is higher than X per sentence of a Y length (golden standard), than rank S–O
Major Ref. Botel and Granowsky (1972) or Szmrecsanyi (2004)
Semantic: use of complex, difficult, technical or unusual terms called 'outliers' (e.g. ‘as necessary’, ‘generally’) or of two or more semantically different CI parameters in information flows
Rank: E.g. if clause contains more outliers than the number set in golden standard, then rank as S–O
ALGO: LOF, CI in information flows
Major Ref. Bartlett et al (2019) Shvartzshnaider et al. (2019)
  Information asymmetry Lack of information Presence of all information required by the law (e.g. identity of data controller, types of personal data collected; goals of treatment, etc.)
Rank E.g. if clause omits more elements than all those necessary according to golden standard, then rank as S–O
Major Ref. Liepina et al (2019) and Costante et al (2012) /or Contissa et al. (2018a, b)
Consistency Internal and External Interaction amongst clauses within the same text and across texts Recurrence of same lexicon and cross reference between different clauses in the same document and across documents
Rank: E.g. if a clause scores lower than the citation network gold standard for cross-reference links or evaluation of textual similarity, then rank as S–O
Major Ref. Panagis et al. (2017) [citation net]; or Nanda et al. (2019) [similarity models]
  1. Ranking: O = Optimal; S–O = sub-optimal
  2. Ranking ought to be done per domain/sector