Many domains use specific terminologies to describe concepts. Being able to explicitly manage such terminologies instead of relying on their common knowledge is beneficial both for newcomers and for people for whom the terminology is not their daily bread and butter. This is especially true for legislative terminologies. We present TermIt, a Semantic Web-based terminology manager that allows domain experts to create and manage high-quality terminologies, link them to normative documents as well as use them to annotate other related documents. We discuss the architecture of the system and the technologies used in its development.
People in diferent domains tend to use specific terminologies to describe concepts. The meaning of these terms can often be in a conflict with the meaning understood by lay folk (an issue in GitHub issue tracking does not necessarily mean a problem with the software), or there may exist subtle diferences in the terms’ semantics. In case of legislation or other normative documents, these subtle diferences may have serious consequences. For example, according to the Czech Ordinance No. 268/2009 Coll (Ordinance on technical requirements for buildings), a building is an above-ground structure, including its underground portion, spatially concentrated and externally enclosed for the most part by perimeter walls and roof construction. On the other hand, according to the Czech Law No. 406/2000 Coll. (Energy Management Act), a building is an above-ground structure and its underground parts, spatially concentrated and externally largely enclosed by external walls and roof structure, in which energy is used to modify the indoor environment for heating or cooling purposes. The diference between these two meanings can be important in case of communication with a public ofice or in a legal dispute.
It is therefore advantageous to be able to manage a terminology of the domain consisting of terms, where each term has a label, i.e. the phrase it is called by and a definition describing the term’s meaning.
TermIt was created to support the creation, management and sharing of such terminologies.
TermIt is a terminology manager based on SKOS (Simple Knowledge Organization System) [
CEUR ceur-ws.org
The terms in TermIt are grouped into glossaries, where each glossary corresponds to a skos:ConceptScheme.1 Each term is a SKOS skos:Concept with basic attributes such as skos:prefLabel, skos:definition, etc. A hierarchy of terms can be built within a glossary (using skos:broader) as well as across glossaries (using skos:broadMatch). Additional non-hierarchical relationships between terms can be specified using skos:related(Match) and skos:exactMatch. Users are also able to create and use additional attributes that are not covered by the built-in data model.
To support use in multilingual use cases (for example, the Czech Standardization Agency uses TermIt to map building construction glossaries in cooperation with other European standardization agencies), all string-based term attributes may contain language-tagged values. The only requirement for a term is to have a label (skos:prefLabel) in the glossary’s configured primary language.
Although domain experts, who are the primary intended users of TermIt, are not expected to do
detailed modeling of the relationships between terms (this task requires an ontology engineer), TermIt
does ofer basic modeling capabilities to more advanced users. One already mentioned is the SKOS-based
ability to create term hierarchies. Another is the ability to classify terms into categories. By default, a
UFO-based [
Data quality is a concern in many cases where people provide the input for an information system
and terminology management is no exception. On the contrary, since the terminology is often based
on normative documents or used in highly regulated conditions, the quality of the terminology is
important. The solution in TermIt is twofold: 1) basic consistency rules, such as the requirement for a
term to have a label that is unique within its glossary, are enforced; 2) a validation service that checks
the quality of terms in a glossary using a (configurable) set of SHACL [
Other terminology management features of TermIt include tracking changes made by users, the ability to create glossary snapshots, comment on terms, and a simple state-based workflow. The user may search for glossaries and terms using full-text search or use faceted search for more fine-grained term exploration. Glossaries may be exported/imported to/from RDF as well as MS Excel.
Terminologies are often based on normative documents. TermIt therefore supports the attachment of documents (possibly consisting of multiple files) to glossaries. These documents can be used as a
seed to populate the glossary based on suggestions from Annotace 2 [
Another use case for document annotation is having an existing glossary and finding the occurrences of its terms in the document. For example, the Prague Institute for Planning and Development maintains multiple glossaries that are then used to provide definitions of terms occurring in the documents it publishes. The Portal of Prague Planning Analytical Materials (PAM)3 is one of the sites where such documents are published. Figure 1 illustrates how the PAM Portal and its maintainers use TermIt.
It is important to note that the document annotation is semi-automatic – the service suggests occurrences of terms, but an authorized user is ultimately required to approve or reject these occurrences. If a new version of a document is uploaded to TermIt, previous occurrences are retained, as long as the changes in the document are not too significant. Annotace expects the content to be HTML (or plain text that it transforms to HTML) and produces RDFa-annotated HTML.4
TermIt as a system consists of the TermIt application and multiple optional services - Figure 2 illustrates
its architecture. The TermIt application itself is designed as a monolith with back-end written in Java
and front-end written in TypeScript. The back-end follows the layered architectural style [
Validation service (as already mentioned in Section 2.1) is used to determine the quality of the term metadata and provide the user feedback as to what they should improve. The service contains a predefined set of SHACL rules and the caller may choose which ones to use for individual validation calls.
Authentication service compatible with the OAuth2 standard can be used instead of the built-in internal user management and authentication mechanism. This configuration is suitable in case the target organization already has an authentication solution in place or when TermIt is to be used with OntoGrapher (see below).
OntoGrapher [
TermIt uses Semantic Web technologies at all levels of its implementation, yet the software libraries it uses ensure that it does not difer from regular Web applications in terms of software engineering and development. The data are stored in a triple store. In particular, GraphDB6 is used for its superior performance and support for custom inference rules. TermIt uses a number of rules, for example, to infer inverse relationships between terms or a term and the glossary to which it belongs. RDF named
3https://uap.iprpraha.cz/en, accessed 2025-07-30 4https://www.w3.org/TR/rdfa-core/, accessed 2025-07-30 5https://sparknlp.org/, accessed 2025-07-30 6https://graphdb.ontotext.com/, accessed 2025-07-30 graphs are used to structure the data – each glossary is stored in a separate named graph (its IRI7 coincides with the glossary’s identifier) and additional named graphs are used for related technical data (change records, comments, etc.).
The repository is accessed using the Java OWL Persistence API (JOPA) [
The front-end is a client-side application written in TypeScript using React. It follows common React application development practices – it uses a state management library, routing, authentication, and authorization workflow – and the user interface (UI) is localized to Czech and English. Since all UI texts are extracted into localization files, adding a translation for another language would be trivial. As mentioned earlier, the front-end communicates with the back-end using JSON-LD with context. From experience, this increases the amount of data transferred for a single object retrieval, but in situations where objects repeat in the content, the amount of data actually decreases significantly, because JB4JSON-LD uses an object’s identifier for its repeated occurrences. jsonld.js, 10 is then used to ensure that all references are replaced with the full object for the rest of the front-end code.
The whole system is published and deployed using the Docker11 containerization platform. This ensures the runtime is the same for all instances and nothing (besides Docker itself) needs to be installed on the host system.
TermIt has been used or is planned to be used by several organizations:
• Prague Institute for Planning and Development
• Czech Standardization Agency
7Internationalized Resource Identifier
8https://spring.io/projects/spring-boot, accessed 2025-07-30
9https://spec.openapis.org/oas/latest.html, accessed 2025-07-30
10https://github.com/digitalbazaar/jsonld.js, accessed 2025-07-30
11https://www.docker.com/, accessed 2025-07-30
• Czech Digital and Information Agency – used in a project of an assembly line of semantic
vocabularies for the Czech legislation [
We have introduced TermIt, a terminology manager based on Semantic Web technologies. We argued that a clear definition of terminology is important, especially in domains where misunderstanding the meaning of a word or a phrase can have an impact on a conversation, a dispute, or a legal act.
Building TermIt with Semantic Web technologies has had a number of benefits. A well established model (SKOS) was used to describe the terms, including their relationships, while retaining the possibility to easily define additional properties. The terms have stable and globally valid identifiers in the form of IRIs, and the terminologies can be published as machine-readable data. On the other hand, some drawbacks have also been discovered – mainly issues with performance for larger amounts of data (glossaries with hundreds or thousands of terms or deep hierarchies).
TermIt can be used in cooperation with additional tools, such as a validation service, a text analysis service, and a tool for modeling detailed relationships between terms.
A more detailed description of TermIt use cases as well as a comparison with other relevant tools can
be found in [
• TermIt back-end at https://github.com/kbss-cvut/termit, • TermIt front-end at https://github.com/kbss-cvut/termit-ui, • TermIt Docker Compose configuration (including installation instructions) at https://github.com/kbss-cvut/termit-docker.
A demo instance of TermIt is available at https://kbss.felk.cvut.cz/termit-demo.