linneo

Standard taxonomy

The taxonomy returned by Taxonomy.standard() (and the base of Taxonomy.load() and TaxonomyBuilder.withStandard()). It is assembled by io.github.devbpm.linneo.internal.StandardTaxonomy, which is in a non-exported package and not part of the public API, from tab-separated resources bundled in the JAR under io/github/devbpm/linneo/internal/.

Contents

	fideslang 3.1.4	Linneo	Total
Data categories	85	40	125
Data uses	56	7	63
Data subjects	15	6	21

The fideslang version is StandardTaxonomy.FIDESLANG_VERSION ("3.1.4"). The fideslang terms are bundled unchanged (license CC BY 4.0, © Ethyca, Inc.; see NOTICE) and have origin fideslang; Linneo’s terms have origin linneo and never redefine a fideslang key. fideslang defines no protection levels: all of them come from Linneo’s policy file.

For every key, use the constants in Categories, DataUses and DataSubjects or iterate Taxonomy.categories(), dataUses() and dataSubjects().

Data category roots

Every root declares all four facets in linneo-category-policy.tsv:

Root	Origin	Nature	Confidentiality	Integrity	Secret	Covers
user	fideslang	ORDINARY	LOW	MEDIUM	no	Data about a natural person
system	fideslang	NONE	LOW	LOW	no	Data of the system: authentication, operations, security, configuration
org	Linneo	NONE	PUBLIC	LOW	no	The organization and third parties that are not natural persons
reference	Linneo	NONE	PUBLIC	MEDIUM	no	Master data from public or standard sources

Protection policy

Below the roots, the policy raises facets where the data requires it; any key not listed inherits from its parent (see Protection levels). The main rules:

Categories	Policy
user.health_and_medical, user.biometric, user.demographic.philosophical_belief, .political_opinion, .race_ethnicity, .religious_belief, .sex_life, .sexual_orientation, .trade_union_membership	SPECIAL_CATEGORY, confidentiality HIGH, integrity HIGH
user.criminal_history	CRIMINAL_OFFENCE, HIGH, HIGH
user.administrative_sanction, user.victim_status	ORDINARY, confidentiality HIGH, integrity HIGH
user.authorization (scope of permissions)	confidentiality HIGH, integrity HIGH; user.authorization.credentials, .password and .biometric are secret, and user.authorization.biometric is also SPECIAL_CATEGORY
user.device.session_id	confidentiality HIGH, secret
system.authentication	confidentiality HIGH, integrity HIGH; system.authentication.secret is secret
user.financial, user.payment, user.education	confidentiality MEDIUM, integrity HIGH; user.financial.credit_card confidentiality HIGH
user.behavior, user.government_id, user.location.precise, user.content.private, user.content.self_image, several user.demographic.* (date of birth, gender, marital status, profile, family, nationality)	confidentiality MEDIUM
user.childrens, user.government_id.immigration	confidentiality HIGH (data of vulnerable data subjects: children and migrants or asylum seekers)
user.behavior.activity_log, system.security, system.integrity	integrity HIGH
system.configuration	integrity MEDIUM
org.structure, org.location, org.process	confidentiality LOW, integrity MEDIUM
org.legal_entity	integrity MEDIUM
org.business_records	confidentiality MEDIUM, integrity MEDIUM

Children of a special-category category are special category too (for example user.health_and_medical.disability), and every descendant of a secret category is secret. The builder requires every SPECIAL_CATEGORY and CRIMINAL_OFFENCE category to resolve to HIGH confidentiality and HIGH integrity (see Protection levels).

Legal basis

The policy follows Regulation (EU) 2016/679 (GDPR) as corrected (OJ L 127, 23.5.2018; EUR-Lex has no later consolidated text) and, for Spain, Ley Orgánica 3/2018 (LOPDGDD), consolidated text of 27 December 2025 (BOE-A-2018-16673):

• Special categories (GDPR and LOPDGDD article 9) and criminal offence data (GDPR and LOPDGDD article 10) get the maximum protection, HIGH and HIGH.
• user.childrens is HIGH because children are vulnerable data subjects (GDPR recital 38 and article 8; LOPDGDD article 7 sets the age of consent at 14).
• user.government_id.immigration is HIGH for the same reason applied to migrants and asylum seekers (EDPB WP248 rev.01).
• Data about administrative infringements and sanctions (user.administrative_sanction) is ORDINARY with high confidentiality and integrity, not CRIMINAL_OFFENCE (LOPDGDD article 27); criminal offences stay in user.criminal_history.
• user.victim_status (recognition as a victim of gender-based violence, sexual violence, terrorism, trafficking or another offence, and the protection and entitlements it gives) is deliberately ORDINARY, not CRIMINAL_OFFENCE, with HIGH confidentiality and integrity.

Linneo’s additions

Data categories (40, all since 0.1.0):

Parent	Added children
user	user.education with program, enrollment, course_history, grade, test_score, previous_education, status, accomplishment; administrative_sanction, victim_status
user.unique_id	record_reference
user.government_id	social_security_number
user.device	user_agent, session_id
user.workplace	team, unit, role, assignment_period
user.demographic	nationality, family, philosophical_belief, trade_union_membership, sex_life
user.financial	scholarship, fee
user.health_and_medical	disability
user.behavior	activity_log
system	system.security, system.integrity, system.configuration
system.authentication	secret
(root)	org with structure, legal_entity, location, process, business_records
(root)	reference with public

Data uses (7): the root education (“Provides educational services”) with education.admission, education.enrollment, education.teaching, education.assessment, education.academic_records and education.financial_aid.

Data subjects (6, all roots): student, applicant, alumnus, faculty_member, staff_member, legal_representative.

fideslang roots

• Data categories: user, system.
• Data uses: analytics, collect, employment, essential, finance, functional, marketing, operations, personalize, sales, third_party_sharing, train_ai_system.
• Data subjects: 15 flat terms, such as customer, employee, job_applicant, patient, visitor.

fideslang 3.1.4 has one deprecated term: the data use personalize.profiling (deprecated in 2.1.1, replaced by personalize.content.profiling). No standard data category is deprecated.

Resource files and formats

File	Origin	Columns
fideslang-data-categories.tsv, fideslang-data-uses.tsv, fideslang-data-subjects.tsv	fideslang	term format (7)
linneo-data-categories.tsv, linneo-data-uses.tsv, linneo-data-subjects.tsv	linneo	term format (7)
linneo-category-policy.tsv	—	policy format (5)

Common rules (UTF-8): blank lines and lines starting with # are ignored; the first remaining line is the column header and is skipped; cells are separated by a tab and stripped of surrounding white space; missing trailing cells are empty; a line with more cells than columns is rejected (TaxonomyException: <file>: too many columns in line '...'). A missing resource throws TaxonomyException (Missing taxonomy resource <file>).

Term format: key, parent_key, name, description, version_added, version_deprecated, replaced_by.

• parent_key must equal the key without its last segment (empty for a root); otherwise TaxonomyException (<file>: '<key>' declares parent '<p>' but its key implies '<q>').
• A non-empty version_added becomes TermInfo.since.
• A non-empty version_deprecated deprecates the term, with replaced_by as replacement (or none if empty). replaced_by is ignored when version_deprecated is empty.
• In the fideslang files an empty version_added is written as 2.0.0, the value fideslang’s default_factory assigns.

Policy format: key, nature, confidentiality, integrity, secret.

• Empty cell = inherit. Non-empty values are enum constant names (ORDINARY, HIGH…); secret is true or false.
• Each row is applied with TaxonomyBuilder.refineCategory after all terms are loaded, so the key must exist.

Loading order: fideslang categories, uses and subjects (origin fideslang), then Linneo’s categories, uses and subjects (origin linneo), then the policy. TaxonomyBuilder.build() then checks the hierarchy and resolves the inherited facets.

Regenerating from fideslang

The fideslang-*.tsv files and the constant classes are generated; do not edit them by hand. From the repository root:

python -I tools/fideslang/update.py 3.1.4

The script (Python standard library only) downloads the release’s source distribution from PyPI, reads src/fideslang/default_taxonomy/data_{categories,uses,subjects}.py and writes:

• src/main/resources/io/github/devbpm/linneo/internal/fideslang-data-{categories,uses,subjects}.tsv, with a header comment naming the source, version and licence;
• src/main/java/io/github/devbpm/linneo/{Categories,DataUses,DataSubjects}.java: the fideslang keys, then the keys of the linneo-*.tsv files (see Key constants).

After adding a Linneo term to a linneo-*.tsv file, rerun the script to regenerate the constants. When moving to another fideslang version, also update StandardTaxonomy.FIDESLANG_VERSION, NOTICE and the term counts asserted in StandardTaxonomyTest, and make sure every new root category has a full row in linneo-category-policy.tsv (otherwise the taxonomy fails to build).

See also: Taxonomy · Protection levels · Key constants · Extending the taxonomy · course: The taxonomy

This site is open source. Improve this page.