A flexible text analyzer based on ontologies: an application for detecting discriminatory language

Abstract

Language can be a tool to marginalize certain groups due to the fact that it may reflect a negative mentality caused by mental barriers or historical delays. In order to prevent misuse of language, several agents have carried out campaigns against discriminatory language, criticizing the use of some terms and phrases. However, there is an important gap in detecting discriminatory text in documents because language is very flexible and, usually, contains hidden features or relations. Furthermore, the adaptation of approaches and methodologies proposed in the literature for text analysis is complex due to the fact that these proposals are too rigid to be adapted to different purposes for which they were intended. The main novelty of the methodology is the use of ontologies to implement the rules that are used by the developed text analyzer, providing a great flexibility for the development of text analyzers and exploiting the ability to infer knowledge of the ontologies. A set of rules for detecting discriminatory language relevant to gender and people with disabilities is also presented in order to show how to extend the functionality of the text analyzer to different discriminatory text areas.

Appendices

Appendix 1: Relevant class descriptions for discriminative language detection

Extra-visibility $\equiv$ DisabledPeople  $\sqcup$ RacePeople $\sqcup$  ReligionPeople $\sqcup$ Sex  $\sqcap$ $\exists$ hasNext Noun

InappropriateTitles $\equiv$ Dr  $\sqcup$ Mr $\sqcap$ $\exists$  hasNext ($\exists$ hasNext (Feminine $\sqcap$  ProperNoun)) $\sqcup$ $\exists$ hasNext  ($\exists$ hasNext ($\exists$ hasNext  (Feminine $\sqcap$ ProperNoun))) $\sqcup$  $\exists$ hasPrevious ($\exists$ hasPrevious  (Feminine $\sqcap$ ProperNoun)) $\sqcup$  $\exists$ hasPrevious ($\exists$ hasPrevious  ($\exists$ hasPrevious (Feminine $\sqcap$  ProperNoun)))  $\sqcup$ Mrs $\sqcup$ Ms $\sqcap$  $\exists$ hasNext ($\exists$ hasNext  (Masculine $\sqcap$ ProperNoun)) $\sqcup$  $\exists$ hasNext ($\exists$ hasNext  ($\exists$ hasNext (Masculine $\sqcap$ ProperNoun)))  $\sqcup$ $\exists$ hasPrevious  ($\exists$ hasPrevious (Masculine $\sqcap$  ProperNoun)) $\sqcup$ $\exists$ hasPrevious  ($\exists$ hasPrevious ($\exists$ hasPrevious  (Masculine $\sqcap$ ProperNoun)))

ManAsVerb $\equiv$ Man $\sqcup$  Manning $\sqcap$ $\exists$ hasNext The

ManPrecededByForInOf $\equiv$ Man $\sqcap$  $\exists$ hasNext (For $\sqcup$  In $\sqcup$ Of)

ManPrecededByForInOf $\sqsubseteq$ ManAlternative 

MenWomenOrder $\equiv$ He  $\sqcap$ $\exists$ hasNext  ($\exists$ hasNext She) $\sqcup$  Him $\sqcap$ $\exists$ hasNext  ($\exists$ hasNext Her) $\sqcup$   His $\sqcap$ $\exists$ hasNext  ($\exists$ hasNext Hers) $\sqcup$  Men $\sqcap$ $\exists$ hasNext  ($\exists$ hasNext Women) $\sqcup$  Sir $\sqcap$ $\exists$ hasNext  ($\exists$ hasNext Madam)

NeutralMasculinePronoun $\equiv$  Masculine $\sqcap$ Pronoun $\sqcap$  $\exists$ isPrecededBy ProperNoun

NeutralMasculinePronoun $\equiv$ Masculine  $\sqcap$ Pronoun $\sqcap$  $\exists$ hasNext ($\exists$  hasNext (Feminine $\sqcap$ Pronoun))

SexistDescription $\equiv$ Adjective  $\sqcap$ $\exists$ hasNext Women  $\sqcap$ $\exists$ hasPrevious  (And $\sqcap$ $\exists$ hasPrevious  (Men $\sqcap$ $\exists$ hasPrevious Adjective))

Stereotyping $\equiv$ Sufferer  $\sqcup$ Victim $\sqcap$ $\exists$  isFollowedBy Illness $\sqcup$ $\exists$  isPrecededBy Illness

Appendix 2: Rules for detecting discriminative language

Rule	Description	Examples
2.1. Extra-visibility	It is quite unnecessary to mention a person’s sex, race, ethnic background, religion or disability	Male nurse; female engineer; muslim student; Black police officer	$\checkmark$
3.1.1. Invisibility	Women are often invisible in language due to the use of the masculine pronouns ‘he’, ‘him’, ‘his’ to refer to both men and women, and the use of ‘man’ as a noun, verb or adjective	Mankind; man made	$\checkmark$
3.1.2. Inferiority	Unnecessary mention of gender to suggest that in certain roles women are inferior to men. The use of ‘feminine’ suffixes such as ‘ette’, ‘ess’, ‘ienne’ and ‘trix’ are unnecessary	Female engineer; woman academic; actress	$\checkmark$
3.2.1. Use alternatives for ‘man’		Mankind; the best man for the job; the man in the street; man of letters, men of science; manpower; manmade	$\checkmark$
3.2.2. Avoid the use of ‘man’ as a verb		We need someone to man the desk; manning the office; She will man the phones	$\checkmark$
3.2.4. Find alternatives to ‘he’ and ‘his’	The pronouns ‘he’, ‘his’ and ‘him’ are frequently used as generic pronouns. As this use is both ambiguous and excludes women, try to find alternatives	The student may exercise his right to appeal	$\checkmark$
3.2.7. Use alternatives for sex-specific occupation terms	Avoid the impression that these positions are male-exclusive. Avoid using occupational titles containing the ‘feminine’ suffixes -ess, -ette, -trix, -ienne.	Chairman; headmaster; headmistress; policeman; businessman; layman; groundsman; actress; executrix; authoress; comedienne	$\checkmark$
3.2.8. Use appropriate titles and other modes of address	The inappropriate use of names, titles, salutations and endearments create the impression that women merit less respect or less serious consideration that men do. Ensure that people’s qualifications are accurately reflected in their title, and that women’s and men’s academic titles are used in a parallel fashion	Albert Einstein and Mrs Mead; Ms Clark and John Howard; Judy Smith and Dr Nguyen	Partially
3.2.9. Use of Ms, Mrs, Miss, Mr	The use of ‘Ms’ is recommended for all women when the parallel ‘Mr’ is applicable, and ‘Ms’ should be used when a woman’s title of preference is unknown		$\checkmark$
3.2.10. Avoid patronising expressions	Use the words ‘man’/‘woman’, ‘girl’/‘boy’, ‘gentleman’/‘lady’ in a parallel manner	The girls in the office; Ladies; My girl will take care of that immediately	$\checkmark$
3.2.12. Avoid sexist descriptions	Avoid the use of stereotyped generalisations about men’s and women’s characters and patterns of behaviour	Strong men and domineering women; assertive men and aggressive women; angry men and hysterical women	$\checkmark$
4.1.1. Derogatory labelling	They are still used, and should be avoided. Some acceptable alternatives for such labels are ’person with Down’s Syndrome’, ‘person with an intellectual disability’	Cripple; mongoloid; deaf and dumb; retarded	$\checkmark$
4.1.2. Depersonalising or impersonal reference	Often people with a disability are referred to collectively as the disabled, the handicapped, the mentally retarded, the blind, the deaf, or paraplegics, spastics, epileptics etc.	The disabled; the handicapped; disabled people; the physically handicapped; a paraplegic; paraplegics; an epileptic; the deaf	$\checkmark$
4.1.3. Stereotyping	Never use the terms ‘victim’ or ‘sufferer’ to refer to a person who has or has had an illness, disease or disability	Victim of AIDS; AIDS sufferer; polio victim	$\checkmark$