History of the Semantic Web Topic Hierarchy
 General motivation
The Semantic Web Topic Hierarchy was created to merge several different collections of topics into a single structure. In summary, it was started from the following activities:
- The European Network of Excellence REWERSE started to develop a graduate curriculum for Semantic Web education in September 2004.
- The European Network of Excellence KnowledgeWeb continued the discussion about the topic hierarchy in June 2005, coming out of the close cooperation with REWERSE in setting up REASE, a repository for Semantic Web learning resources, which uses the topic hierarchy as a basis for the REASE catalogue.
- The International series of Semantic Web conferences (ISWC) and the European series (ESWC) was using a couple of topics for the classification of their papers into categories, which were used for the assignment of paper submissions to reviewers as well as for naming sessions during the conference.
- In May 2007, KnowledgeWeb decided to use this wiki as main dissemination channel for the project results.
Of course, the topic hierarchy is by no means perfect and reflects the subjective opinion of the involved researchers about the field 'Semantic Web'. It has undergone several changes and versions, as more people started to contribute to the discussion of the topic hierarchy and also because the 'Semantic Web' research area developed over time during the years 2004-2007, when the topic hierarchy was initiated.
This document is intended to capture the rationales for the specific changes made to the topic hierachy and provide pointers to the documents where the changes were introduced
 History of Deliverables in REWERSE and KnowledgeWeb relevant for the topic hierarchy
 REWERSE E-D1: Collecting existing university courses about Semantic Web topics (Sept. 2004)
To initiate the discussion about developing a Semantic Web graduate curriculum, existing university courses about Semantic Web topics were collected jointly by KnowledgeWeb and REWERSE in the first months of 2004 (cf. REWERSE deliverable E-D1). As a result, four main categories were identified:
- Foundations (relevant to Semantic Web), such as databases, logics, knowledge representations, and logic programming
- Basic Web Information technology, such as XML, web data integration
- Semantic Web concepts and techniques, such as RDF, OWL, and rules
- Other related techniques, such as agents, security, business rules, or knowledge management
It became clear that a better structuring of the field of Semantic Web is required for a final Semantic Web curriculum.
 REWERSE E-D5: REWERSE Graduate courses (Feb. 2005)
- IS (intelligent systems)
- IS3 Knowledge Representation and Reasoning (core)
- IS6 Agents (elective)
- IS7 Natural Language Processing (elective)
- IM (information management)
- IM2 Database systems (core)
- IM3 Data modelling (core)
- IM12 Hypertext and Hypermedia (elective)
However, many important concepts such as basic web information technologies and the specific Semantic Web concepts were missing in the ACM curriculum. Hence, the following preliminary structure of the topic hierarchy was presented (listed here because it's not available in the history of the page for the Semantic Web Topic Hierarchy.
- IS Intelligent Systems
- IS1 Knowledge Representation and Reasoning
- Logics (Predicate Logic, Description Logics, Horn Logic, F-logic, Modal Logics)
- Basics of automated reasoning
- Logic Programming and Non-monotonic Reasoning
- Reasoning on Action and Change
- Temporal and Spatial Reasoning
- IS2 Agents
- IS3 Natural Language Processing
- IS1 Knowledge Representation and Reasoning
- IM Information Management
- IM1 Data Modeling
- Conceptual models; ontologies, UML
- Relational data model
- Semistructured data
- Object-oriented model
- IM2 Database systems
- Relational databases
- Object-oriented databases
- Distributed databases
- IM3 Hypertext and Hypermedia
- IM1 Data Modeling
- WT Basic Web information technologies
- WT1 XML
- Schema languages
- XML query and transformation languages
- XML programming techniques
- WT2 Web data integration
- WT3 Security
- WT4 Web services
- WT5 Personalization techniques
- WT1 XML
- SW Semantic Web
- SW1 Basic ideas of the Semantic Web
- SW2 Resource Description Framework
- RDF basics
- RDF Schema
- RDF Query Languages
- SW3 Ontologies
- Web Ontology languages
- Ontology Engineering
- SW4 Rules on the Semantic Web
- SW5 Semantic Web Applications
- Web Services
 KnowledgeWeb deliverable D3.1.5: REASE
Starting from E-D5, a discussion was initiated on the KnowledgeWeb education area mailing list and the REWERSE education and training mailing list in June 2005 with the special focus to extend the original REASE catalogue from September 2004 as described in the KnowledgeWeb deliverable D3.3.2v2. As a result, version 1.0 of the Semantic Web Topic Hierarchy was created extending the previous proposal with several new main topics and refining existing topics. Specifically, it included also the main session topics from previous International and European Semantic Web Conferences (ISWC and ESWC), the work packages of KnowledgeWeb and the topics of the KnowledgeWeb summer school (the topics of the REWERSE summer school were already included in E-D5).
Version 1.0 of the topic hierarchy was reported in the KnowledgeWeb deliverable D3.1.5 in December 2005 and announced on the KnowledgeWeb and REWERSE mailing lists in January 2006. It was also used to describe the current state of REASE in REWERSE deliverable E-D6 in August 2005 after a first draft of it has been used in REASE in August 2005.
The main changes were as follows:
- Copying Knowledge Engineering / Ontology Engineering to the foundations (from SW) as it has existed long before the Semantic Web came to life; Semantic Web specific issues around ontology engineering are supposed to be classified into the SW3 parts while the Semantic Web independent parts are now to be classified into the foundational category 'Knowledge Engineering / Ontology Engineering'.
- The reasoning part in the foundations was cleaned
- Within Basic Web Information Technologies two categories were added: Web Data Extraction and Architecture of Web Information Systems
- The Semantic Web main category was extended along the Semantic Web layer cake, adding three new categories:
- Semantic Web Infrastructure
- Semantic Web Proof
- Security / Trust / Privacy
- The main category 'Semantic Web Special Topics' was introduced. It has been in the initial version of the REASE catalogue and was now merged with the 'other related areas' part. For duplicated categories like Natural Language Processing, the main idea again was to put the Semantic Web specific things into the 'special topic' category while the Semantic Web independent parts are classified under 'IM Natural Language Processing' in the foundational categories.
 REWERSE E-D7: Graduate Education Curriculum
REWERSE deliverable E-D7 reported in February 2007 about version 1.1 of the topic hierarchy, which included some minor modifications as a result of the analysis of existing material in REASE and learning resources for the Semantic Web curriculum as well as including feedback from the first public announcement.
- Added Temporal Logics as it is very important for example for the Agents community with the Semantic Web community.
- Removing `Information Management' (originally included from the ACM CCS) because it was too general.
- Removing `Semantic Web Infrastructure' because it turned out that Semantic Web Services are actually similar to the many topics in `Semantic Web Special Topics' and the remaining `Architecture' did not really justify its own main topic (it was actually very broad itself and never used for any categorization of resources in REASE, for example).
- Replacing `Reasoning Engines' (this category was too general under the super-topic `Rules') by Ontology Reasoners and Rule Reasoners
- Adding Integration of Rules and Ontologies
- Adding eGovernment as an important application domain for Semantic Web technologies.
- Adding Design and Testbed Case Studies.
As the ontoworld wiki supports `subtopic' links, we integrated the topic hierarchy into the category system of the wiki. So there is one wiki page for each topic which provides a brief explanation of the topic plus an automatically generated overview about all other wiki pages which have been annotated with this topic. Furthermore, we provide links to find resources about this topic (e.g., to REASE).
As an example, for the topic `ontology engineering' the wiki contains 21 subcategories and 9 plain wiki pages being tagged with this topic (cf. Ontology Engineering).
Version 1.1 was the first version where all topics are described by separate 'Topic' pages in the wiki.
 KnowledgeWeb D3.1.5v2 and REWERSE E-D9: Evaluation of the topic hierarchy
- manual inspection by an expert not yet being involved in the definition
- comparison to existing author keywords in the DBLP++ corpus
- comparison to automatically generated topic hierarchies using the Semantic GrowBag approach
As a result, several changes were proposed for inclusion in version 2 of the topic hierarchy:
- Add Information Retrieval / Search (actually being IM11 in ACM CC)
- Add Machine learning (actually being IS8 in ACM CC)
- Add 'document classification' (which was finally rejected because document classification is often used to generate Semantic Web content (=metadata), but there is no Semantic Web specific part in document classification).
 Version 2 of the topic hierarchy
Version 2 of the topic hierarchy was based upon the following information:
- Collecting all feedback about previous version of the topic hierarchy (e.g., from the development of the joint master curriculum or the classification of learning materials in REASE, or from the creation of the KnowledgeWeb technology roadmap).
- Integrating topics which were added by other people using the Ontoworld wiki (a true community effort)
For all topics, their occurrence in google scholar, google, and FacetedDBLP were analyzed to ensure that they are somehow 'commonly used'. As a result, the main structure of version 1.1 of the taxonomy remained unchanged and consists of Foundations, Semantic Web Core Topics and Semantic Web Special Topics. In general, we have tried to simplify the topic hierarchy and to clean duplicates as far as possible, as well as to add categories only if we had strong indications of their importance. Furthermore, we have opted to delete categories for which we could not see any strong indication that they are actually important for the field (initially, we included basically all proposed areas as we assumed that areas that are worked upon in KnowledgeWeb / REWERSE are also of a somewhat importance).
Hence, the changes compared to version 1.1 are related to obsolete categories, changed categories, and additional categories as explained in the following sections.
- Renamed all subcategories of ontology engineering to include 'ontology' for practical reasons (like searching for related resources on external sources like google scholar).
- Refining category Ontology Population / Generation to Ontology Construction / Ontology Learning / Ontology Population as ontology learning is by far the most prominent term and ontology generation is used rather rarely.
- Added Ontology Evolution as this is a popular term to express dynamics similar to (but more popular as) Maintenance and Versioning of ontologies (this term was introduced by other users of the ontoworld wiki and subsequently integrated).
- Added Ontology Merging as this also is a very popular term / synonym for Ontology Integration.
- Added Ontology Visualization as requested by several people
- Removed Predicate Logics as it is covered by first-order logics
- Removed Modal Logics as it is not so important for the Semantic Web and the topic hierarchy should do fine with the other Logics categories.
- Removed the subtopics of 'XML' since 'namespaces' and 'programming techniques' are too detailed, 'query languages' are now available in a different category, and 'schema languages' did not justify another hierarchical level and is also a bit too detailed here)
- Added 'Fuzzy Logic'.
- `Reasoning' refined by `Reasoning Engines / Theorem Provers' and `Fuzzy Reasoning'
- Added Web service discovery and Web service composition as other people already used them in ontoworld and they are indeed important subtopics of 'web service'.
- Removed Architecture of Web Information systems as it was too general and not really used
- Moved Security as a general security category and moved Security / Trust / Privacy to emerging topics (not too much Semantic Web specific material there for now even though it has been part of the Semantic Web layer cake for ages...)
- Added Rules and their sub-topics Deductive Rules, Reactive Rules, Rule Visualization to the foundations as suggested by Francois Bry (rules have been around before the Semantic Web just as ontology engineering has) and left only the SW-specific parts in 'Semantic Web rules'
- Added main foundational category Information Access, comprising the subtopics Query languages, Browsing / Navigational Access, Query Algebra, Visual Querying, and Event queries as also suggested by Francois Bry to also separate the 'non-Semantic Web specific' query things from the Semantic Web specific ones.
 Semantic Web Core
- Renamed 'Semantic Web Query and Update Languages' to 'Semantic Web Information Access' to make it consistent with 'Information Access' in general
- Dropped 'Semantic Web Update Languages' as there is not much about it yet (moved it to Emerging Semantic Web Topics).
- Added Semantic Web Browsing as it is a popular research area and several prototypes have been developed recently.
- Refined Ontologies in the Semantic Web:
- Added a separate category OWL because it's one of the main components of the Semantic Web
- Moved Resource Description Framework / RDFSchema to here as its also an ontology language
- Removed 'ontology engineering' since there is no Semantic Web specific ontology engineering for now and the rest is covered by the foundation category 'ontology engineering'
- Removed 'ontology reasoners' since it is not a commonly used term
- Added Legacy Ontology Languages (DAML, DAML+OIL) to be able to refer to the foundations of the Semantic Web languages
- Added Ontology Repositories as there is a lot of work going on here
- Added Ontology Instances to be able to find examples of ontologies
- Added Upper-Level Ontologies / Top-Level Ontologies as the former was already used in the ontoworld wiki and the latter is a similarly often used synonym
- Added Domain Ontologies as this was already used in the ontoworld wiki
- Renamed 'Rules + Logics' to Rules as logics are already covered by the foundational part
- Removed 'Rule reasoners' as it is not a commonly used term
- Added Distributed rule processing as it is an important topic within 'rules for the semantic web'
- Removed 'Semantic Web proof' since the term is not commonly used and Semantic Web reasoning is covered by other categories.
- Removed 'Semantic Web security / privacy / trust' as apparently not many people are currently working on this topic
- Rename 'Applications' to Application domains and
- Added several application areas: Information Retrieval / Search (as an outcome of analyzing cooccurring keywords in SW publications), e-Culture, Human resources, Blogs, Business Rules (already in E-D1, but not considered initially), Wikis, Digital Libraries (IM14 in the ACM CC2001), Data Integration
- Added Reasoning in the Semantic Web for the categorization of Semantic Web specific issues.
 Semantic Web Special Topics
- Removed 'design and test bed case studies' again as it did not attract any attention in the previous version.
- Added Semantic Desktop