This paper describes the features required of OWL to realise and enhance Indigenous Knowledge (IK) digital repositories. Several needs for Indigenous Knowledge management systems (IKMSs) are articulated, based on extensive stakeholder input, and analysed on the suitability of semantic web technologies in addressing them. Based on their potential for impact and maturity, several possible applications are recommended for further investigation and inclusion into current or new IKMSs, including: ontology based querying and browsing; a natural language independent ontology for multilingual data access; support for collaborative knowledge generation; and the formalisation of IK for scienti c discovery. For each of these possible applications, the required OWL features are discussed, which include representation of vagueness, mereotopology, modularisation, and extended support for internationalisation and annotation.
Indigenous Knowledge (IK) is the local, traditional knowledge held by people of a
particular area. It is central to their cultural heritage and holds signi cant value.
The collection and management of IK is increasingly important for the purposes
of preservation, protection, conservation and promotion [
While the Semantic Web is still a work in progress, innovative tools and technologies have been developed that can be utilised in software development today. In particular, there are opportunities for utilising ontologies and semantic web technologies in the area of Indigenous Knowledge management systems (IKMSs) in order to address some of the challenges in management and dissemination of IK. For example, collected IK is often unstructured, transferred in di erent local languages and described using local vernacular terms. This creates unique challenges that may be e ectively addressed through the use of semantic web technologies.
The aim of this paper is to summarise the outcome of our exploration of existing semantic web technologies and the opportunities they create when applied in the domain of IK management. This semantic web technologies-based requirements analysis is subsequently used to formulate the speci c OWL features required for them to be e ective in IKMSs. We focus on four of the high-level speci ed needs and link them to eight areas of semantic web technologies. The four needs are ontology-based data access, automated reasoning for scienti c knowledge discovery, multilingual ontologies, and collaborative ontology development. The former two result in requirements for extensions to the standard OWL with respect to the expressiveness of the underlying logic, whereas the latter two concern other OWL features that would help realising development and use in real systems.
The paper is structured as follows. The IKMS needs are summarised in Section 2. The survey of applicable semantic web technologies is presented in Section 3. Selected opportunities for application in the domain of IKMSs are described in Section 4 with a focus on the OWL features required for them to be e ective. The paper is concluded in Section 5. 2
IK
management system needs IK can be de ned as the traditional knowledge that is unique within a community or society, transferred by sharing experiences and skills and by storytelling from generation to generation. IK is a very wide eld and spans a multitude of themes such as story telling; food technology; healing and nutrition; arts and crafts; cosmology; and traditional medicines.
Since IK is primarily implicit, it is in danger of being lost unless it is captured and preserved. IKMSs are formalised systems, often supported through technology, with the primary purpose of collecting, managing, preserving and disseminating IK. Collected IK is often unstructured, transferred in di erent local languages and described using local vernacular terms. This creates unique challenges that may be e ectively addressed through the use of semantic web technologies.
The needs for IKMSs were explored through workshops conducted by the authors with potential users of IKMSs in South Africa. The participants included representatives from government departments, higher education institutions, science councils, scientists, traditional authorities and community based organisations. The aspects explored during these workshops included the questions to be answered on IK, the information on IK to be preserved and the functionality required for e ective IK management, preservation and dissemination. The results from the workshops were prioritised based on input from the participants and ltered to yield the following list of needs that could potentially be addressed by semantic web technologies: 1. E ective interrogation of IK: The unstructured nature of IK and the fact that it often contains vernacular concepts makes interrogation and dissemination extremely challenging. In particular, the ability to query IK, browse IK based on inherent structures and relationships and nd answers to complex questions were identi ed as speci c needs. 2. Access to multi-lingual information: IK is often collected in the native language of the IK holder. Automatic translation services are not yet readily available for all African languages and manual translation is expensive and time consuming. The ability to access and query over information in di erent languages will be very bene cial in the context of IK management. 3. Collaborative knowledge generation: IK is a shared resource over individuals in traditional communities, geographical areas or communities of practice. A facility to capture and annotate the knowledge in a collaborative fashion will increase the e ectiveness of IK collection and management initiatives. 4. Information classi cation: Due to the unstructured nature of IK it is difcult to classify the collected knowledge correctly. Classi cation is required to e ectively structure IK repositories. 5. Formalisation of information: IK is a valuable resource for responsible scienti c discovery. The formalisation of applicable IK will enhance the effectiveness of scienti c exploration. 3
Based on the selected needs as described in the previous section, a number of
existing semantic web technologies that may be applicable in IKMSs were
identi ed. They were analysed based on their maturity and application readiness in
IKMSs and four were selected as appropriate for implementation [
The technologies considered most relevant for application in IKMSs are described below, together with a short explanation of their relevance: Semantically enhanced querying uses semantics in search, resulting in more precise results. Reasoning services over ontologies also allow the system to compensate for incomplete information. Ontology-based search is an emerging eld, with several prototypes but few scalable solutions. It can be of particular value in IKMSs, as it can allow sophisticated and accurate querying over complex knowledge structures within IK repositories.
Semantic browsing of information allows browsing of information in a system based on the concepts and relationships de ned in an ontology. Tools for ontology-driven navigation are mature; however, further work is needed for robust and scalable linkages to underlying data sources. Semantic browsing of collected IK can be particularly powerful as it will facilitate guided navigation through the complex knowledge structures within IK repositories.
enable scienti c veri cation and exploration of captured IK. The technology is still immature but the impact of success will be high.
Multi-lingual access to information using ontologies enables browsing and searching of knowledge in di erent languages and of accessing related information stored in di erent languages. This eld is mature with a number of successful applications and can be particularly powerful in promoting navigation of recorded IK by indigenous communities in their own languages.
nities to create precise representations of their IK in their area of interest in a collaborative manner, using tagging and metadata. This technology is mature with a number of robust applications.
In order to promote the adoption of these technologies, they were studied further, with a focus on the OWL features required to e ectively apply them in IKMSs. The selected technologies are discussed in the following section. 4
We rst discuss two main areas that impact on OWL expressiveness directly, namely semantically enhanced querying and browsing of information, and scienti c knowledge discovery, and then on two that focus on usability extensions with a potential for logic-based extensions, namely multilingualism, and collaborative ontology development. Each topic has a brief description and is followed by the OWL requirements. 4.1
Description Semantic web technologies can enable enhanced querying of IK, by going beyond the simple string matching used in keyword-based search and using the semantics of the metadata stored with the IK. Keyword searches use string and linguistic matching of the search phrase and the information to be searched, but cannot exploit subject domain semantics and knowledge about the structure of information to nd better results. Using ontologies allows the use of semantics in search, resulting in more precise and relevant results, even across institutional/software system boundaries.
Example 1. Plant A is used to treat lung conditions and plant B to treat shortness of breath. Searching for treatments for lung conditions in an ontology-driven system uses the relationship between breathing and lungs to return both A and B, while a standard keyword query will return only A. }
Using reasoning services over ontologies also allows the system to compensate for missing (incomplete) information during the execution of a query. Thus, even if not all the information was explicitly captured in the system, it can, through inference, deduce the correct answer to a query. This is not possible in standard queries only over relational databases.
Example 2. The tuber of the protected plant Disa polygonoides (i.e., orchid or Uklamkleshe) is used to treat voice loss after illness. It is captured in the system that Disa polygonoides occurs in the KwaZulu-Natal province, but it is not captured explicitly where the tuber occurs. A query can, through reasoning, nd the habitat of the tuber, even though it was not explicitly stated, by inferring knowledge from parthood (of the plant), and mereotopological and spatial theories (to deduce the location).
The fruits of Eugenia albanensis (i.e, Vlakappel or Umnanjwa) are eaten to treat diarrhoea. As above, the location of the fruits can be deduced from that of the plant. }
We note that an ontology is a formal representation of the knowledge from domain expert's point of view. It is thus a logical theory that contains knowledge represented according to the domain experts' understanding. This creates the opportunity to have a facility to browse the information in the system based on the concepts de ned in a domain and the relationships between them. Ontologyguided navigation will enable users to discover information based on the meaning of the topic and its relationship with other topics in the domain.
Examples of current research and proofs of concept of this approach include:
Ontology-Based Data Access [
Representing and reasoning over spatial knowledge and the objects occupying those regions of space (mereotopology) is known to be di cult with OWL, in particular regarding the properties of the relations (re exivity, symmetry, etc.), the limitations for the so-called composition tables (arbitrary role composition), and scalability with a large number of instances [11, 12]. The informal example with the tuber amounts to a TuberX v 9properPartOf:P, Trans(properPartOf), PlantY v 9locatedIn:HabitatZ where HabitatZ is a habitat type (say, Wetlands) that is realised in some particular geographical area (say, the eThekwini bay) where the plant grows, then we can assert properPartOf locatedIn v locatedIn in OWL 2 DL, so that we infer that TuberX is also located in HabitatZ. In a second step of the query, one can retrieve the instances of HabitatZ, but this does not link PlantY to the eThekwini bay in particular and there may be wetlands that do not have PlantY, which we are not interested in and should not be in the answer. Put di erently, it seems we have to link type-level knowledge with instancelevel data, for which one could use punning features and be not fussy about the ontological status of species. However, with many geographic areas and species, this becomes cumbersome to maintain and is not very scalable when taking into account the object property characteristics and property chains. Perhaps a language with only simple property chains, transitivity, and punning may have nicer computational properties for large ABoxes than OWL 2 DL.
The second requirement is yet to be worked out in more detail. Our rst assessment leans strongly toward rough and fuzzy extensions to OWL and its extended automated reasoning services (e.g., [13, 14]). Roughness can be useful to analyse the individuals in the knowledge base so as to re ne the classes in the ontology in a bottom-up fashion [15], yet this is still a manual task and the rough subsumption reasoning of [13] is yet to be implemented. Moreover, because one cannot even have both the basic semantics of roughness and scalability of implementations [15], some sort of software-supported, intelligent, dynamic, linking and `conversion' system between OWL 2 DL and its OWL 2 QL pro le is needed. Fuzziness is useful in several scenarios, covering both the knowledge representation and some information retrieval requirements. For instance, it is useful to have fuzzy concrete domains with fuzzy data types, fuzzy concepts, and to use a degree of truth for some axiom [14]. Examples of this include: some plant can be found nearby a pond or nearby another that is nearby a pond; peri-peri (chillies) are grouped into four types; and the perceived e ectiveness to a degree of using plant X for ailment Y, respectively. Further, for artefact annotation and retrieval, DL-Media [10] may be useful; although it is based on DLR-Lite only (cf. OWL 2 in [14]), it is expected to be implemented in a separate subsystem of the IKMS. 4.2
Description Formalising the knowledge and information captured in the system can enable scienti c veri cation and discovery. The IK captured in the system can be formalised into an ontology and enriched with scienti c information. This will allow scientists to use the knowledge for purposes such as hypothesis testing and consistency testing of theories.
In order to utilise the full power of ontologies, new information received by the system must be accurately classi ed according to the de ned concepts in the knowledge base. The accurate classi cation of information will enhance the comprehensibility of the knowledge and the accessibility and ease of retrieval of the information.
Examples of current research and technology include: Taxonomic classi
cation: the classi cation of knowledge on a class level based on the declared
properties in the knowledge base; and Formal concept analysis: using a collection
of objects and their properties to automatically derive an ontology or extend
an existing knowledge base using the knowledge base itself together with
information provided by a domain experts [16]. Proofs of concepts and applications
available include: an automatic system for addressing the Chemical Compound
problem, by interpreting transformations on the compound structures as
updates in an ontology [17]; automated reasoning services for bio-informatics [18];
hypothesis testing using rough ontologies [15]; testing the di erences of versions
of a knowledge base using semantic di [19]; and automatic classi cation of
protein phosphates using an ontology, resulting in classi cation that surpassed
that of human annotators and identi ed gaps in the theory that would not have
been possible otherwise [
Description Ontologies are logical representations of a domain and can thus be natural language independent. This creates the opportunity to enable browsing and searching of knowledge in di erent languages and of accessing related information stored in di erent languages.
Example 3. A query can be formulated in any supported natural language (e.g. any of the 11 o cial languages in South Africa). Internally, the query is translated into a query over the ontology, which is language-independent, and results can be extracted and presented in any of the available languages. }
Examples of current research and technology include: query expansion for
queries in di erent languages [
MesoscopicSmall : Natural ! [0; 1] as a fuzzy datatype,
MesoscopicSmall(x) = trz(x; 1; 5; 13; 20) with trz the trapezoidal function, so that we can have
SmallSpear Spear u 9size:MesoscopicSmall and subsequently declare something alike
Ingcula SmallSpear u 9hasShape:Bladed u 9participatesIn:Hunting.
This is actually just one of the possibilities of a formalised translation of an
English natural language description, not a de nition of ingcula as it may
appear in an ontology about IK of hunting. But let us assume for now we do
want to go in this direction, then we require more advanced capabilities than
even lexicalised ontologies, which only link dictionaries and grammars to the
ontologies (for most South African languages, dictionaries are limited in size
and soft copy availability). A possible solution to this impasse is to use
"connections between natural language-dependent versions either by connecting,
e.g., inqina.owl#Ingcula@zu to a set of axioms or to a dummy class (say,
hunting.owl#SmallBladedHuntingSpear@en) with the above-mentioned de nition for
Ingcula. The current integration of "-connections with OWL [
One could argue about whether a speci cation for inter-ontology mappings should be part of the OWL standard, and we are trying various alternatives, but given the recent activities and the various approaches toward multilingual ontologies, some coherent, interoperable, framework for multilingualism and OWL would be a welcome addition. 4.4
Description Semantic web technologies can be utilised to provide a facility to enable a community to create a precise representation of the knowledge in their area of interest in a collaborative manner. Content loaded can be tagged by the community to enrich the meaning and accessibility of the items and inform the de nition of their metadata for that area of interest.
Example 4. A community of drum builders can collaborate to de ne the domain of their interest by collaboratively developing an ontology or conceptual data model of the domain. As content is added to the system, community members can use this shared representation to tag and annotate the content. }
Current research and technology include: Modelling tools such as mind maps,
graphs, conceptual models that provide an intermediate representation,
collaborative ontology development tools with semantic wikis [
Collaborative ontology development tools exist, as well as basic implemented
migration paths from intermediate representations such as mind maps or the
Semantic Wiki MoKi [
Several important needs for indigenous knowledge management systems were described and analysed to identify where ontologies are relevant. Four areas were selected for analysis and requirements speci cation with respect to OWL and its technical infrastructure for automated reasoning, being (1) enhanced, ontology based querying and ontology navigation to browse the knowledge in the IKMS; (2) formalisation of scienti c knowledge and discovery; (3) a language independent ontology for multilingual data access; and (4) a facility to support collaborative knowledge generation. While some requirements can be met with the existing OWL standard, improvements will be useful regarding the interaction of more and less expressive OWL species, fuzzy ontologies, better support for a multilingual setting (including enhanced "-connections and linking classes to very small modules), and annotation-dependent reasoning.
Currently, we are experimenting with isiZulu verbalisations of OWL ontologies and exploring semantic annotations of digitised cartographic maps with mereotopological relations. Ongoing and further research will entail a prioritisation of the possible applications and the development of a demonstration case in the National Recordal System to practically show the value of ontologies in IKMSs.
Acknowledgements This work was made possible with the support of the South African Department of Science & Technology's National IK Systems O ce (RA, TF) and the EU's FP 7 Net2 project, which funded a collaboration visit from the Free University of Bozen-Bolzano to Meraka in South Africa (CMK). Int. Reasoning Web Summer School (RW 2009). Volume 5689 of LNCS., Springer (2009) 255{356 4. Rodriguez-Muro, M., Lubyte, L., Calvanese, D.: Realizing ontology based data access: A plug-in for Protege. In: Proceedings of the Workshop on Information Integration Methods, Architectures, and Systems (IIMAS 2008), IEEE Computer Society Press, pp. 286289. (2008) 5. Miles, A., Bechhofer, S.: SKOS Simple Knowledge Organization System Reference.
W3c recommendation, World Wide Web Consortium (W3C) (2009) 6. Calvanese, D., Keet, C.M., Nutt, W., Rodr guez-Muro, M., Stefanoni, G.: Webbased graphical querying of databases through an ontology: the WONDER system. In: Proc. of ACM SAC'10, ACM (2010) 1389{1396 March 22-26 2010, Sierre, Switzerland. 7. Franconi, E., Guagliardo, P., Trevisan, M.: An intelligent query interface based on ontology navigation. In: Workshop on Visual Interfaces to the Social and Semantic Web (VISSW2010), IUI2010. (2010) 8. Doms, A., Schroeder, M.: GoPubMed: exploring PubMed with the gene ontology.
Nucleic Acids Research 33 (2005) W783 { W786 9. Muller, H., Kenny, E., Sternberg, P.: Textpresso: An ontology-based information retrieval and extraction system for biological literature. PLoS Biology 2(11) (2004) e309 10. Straccia, U., Visco, G.: DLMedia: an ontology mediated multimedia information retrieval system. In: Proceedings of the International Workshop on Uncertainty Reasoning for the Semantic Web (URSW-08). (2008) 11. Grutter, R., Bauer-Messmer, B.: Combining OWL with RCC for spatioterminological reasoning on environmental data. In: Proc. of OWLED'07. (2007) 6-7 June 2007, Innsbruck, Austria. 12. Keet, C.M., Fernandez-Reyes, F.C., Morales-Gonzalez, A.: Representing mereotopological relations in OWL ontologies with ontoparts. In Simperl, E., et al., eds.: Proc. of ESWC'12. LNCS, Springer (2012) 13. Keet, C.M.: Rough subsumption reasoning with rOWL. In: Proceeding of the SAICSIT Annual Research Conference 2011 (SAICSIT'11), ACM Conference Proceedings (2011) 133{140 Cape Town, South Africa, October 3-5, 2011. 14. Bobillo, F., Straccia, U.: Fuzzy ontology representation using OWL 2. International
Journal of Approximate Reasoning 52 (2011) 1073{1094 15. Keet, C.M.: On the feasibility of description logic knowledge bases with rough concepts and vague instances. In: Proc. of DL'10. CEUR-WS (2010) 314{324 4-7 May 2010, Waterloo, Canada. 16. Baader, F., Ganter, B., Sertkaya, B., Sattler, U.: Completing description logic knowledge bases using formal concept analysis. In: Proc. of IJCAI'07. (2007) 230{ 235 Hyderabad, India, January 6-12, 2007. 17. Mosca, A., Palmonari, M.: Action based ABox update: an example from the chemical compound formulation. In: Second Formal Ontologies Meet Industry Workshop (FOMI 2006), 14-15 December 2006, Trento (Italy). (2006) 18. Keet, C.M., Roos, M., Marshall, M.S.: A survey of requirements for automated reasoning services for bio-ontologies in OWL. In: Proceedings of the 3rd Workshop on OWL: Experiences and Directions (OWLED 2007). Volume 258 of CEUR-WS. (2007) 6-7 June 2007, Innsbruck, Austria. 19. Franconi, E., Meyer, T., Varzinczak, I.: Semantic di as the basis for knowledge base versioning. In: Proc. of NMR'10. (2010) 14-16 May 2010, Toronto, Canada.