This paper outlines a project (involving FAO, SIFAR, and CNR) aimed at building an ontology in the fishery domain. The ontology will support semantic interoperability among existing fishery information systems and will enhance information extraction and text marking, envisaging a fishery semantic web. The ontology is being built through the conceptual integration and merging of existing fishery terminologies, thesauri, reference tables, and topic trees. Integration and merging are shown to benefit from the methods and tools of formal ontology.
1.1
Specialized distributed systems are the reality of today’s information systems architecture. Developing specialized information systems/resources in response to specific user needs and/or area of specialization has its own advantage in fulfilling the information needs of target users. However, such systems usually use different knowledge organization tools such as vocabularies, taxonomies and classification systems to manage and organize information. Although the practice of using knowledge organization tools to support document tagging (thesaurus-based indexing) and information retrieval (thesaurus-based search) improves the functions of a particular information system, it is leading to the problem of integrating information from different sources due to lack of semantic interoperability that exists among knowledge organization tools used in different information systems.
The different fishery information systems and portals that provide access to fishery information resources are one example of such scenario. This paper demonstrates the proposed solution to solve the problem of information integration in fishery information systems. The proposal shows how a fishery ontology that integrates the different thesauri and taxonomies in the fishery domain could help in integrating information from different sources be it for a simple one-access portal or a sophisticated web services application.
Fishery Ontology Service (FOS) is a key feature of the Enhanced Online Multilingual Fishery Thesaurus, a project aimed at information integration in the fishery domain. It undertakes the problem of accessing and/or integrating fishery information that is already partly accessible from dedicated portals and other web services.
The organisations involved in the project are: FAO Fisheries Department
(FIGIS), ASFA Secretariat, FAO WAICENT (GIL), the oneFish service of SIFAR,
and the Ontology and Conceptual Modelling Group at ISTC-CNR. The systems to be
integrated are: the "reference tables" underlying the FIGIS portal [
The official task of the project is "to achieve better indexing and retrieval of
information, and increased interaction and knowledge sharing within the fishery
community". The focus is therefore on tasks (indexing, retrieval, and sharing of
mainly documentary resources) that involve recognising an internal structure in the
content of texts (documents, web sites, etc.). Within the semantic web community
and the intelligent information integration research area (cf. [
In this paper we sketch an outline of the FOS development and some hint of the functionalities that it carries out.
An example of how formal ontologies can be relevant for fishery information services is shown by the information that someone could get if interested in aquaculture.
In fact, beyond simple keyword-based searching, searches based on tagged content or sophisticated natural-language techniques require some conceptual structuring of the linguistic content of texts. The four systems concerned by this project provide this structure in very different ways and with different conceptual AQUACULTURE uf aquiculture uf mariculture uf sea ranching NT1 fish culture
NT2 fish feeding NT1 frog culture rt agripisciculture rt aquaculture equipment Fr aquaculture
Es acuicultura AQUACULTURE uf Aquaculture industry uf Aquatic agriculture uf Aquiculture NT Brackishwater aquaculture NT Freshwater aquaculture NT Marine aquaculture rt Aquaculture development rt Aquaculture economics rt Aquaculture engineering rt Aquaculture facilities
The AGROVOC thesaurus seems to frame aquaculture from the viewpoint of techniques and species. On the other hand, the ASFA aquaculture hierarchy is substantially different: ’textures’. For example, the AGROVOC and ASFA thesauri put aquaculture in the context of different thesaurus hierarchies; an excerpt of the AGROVOC result is (uf means used for, NT means narrower than; rt means related term, Fr and Es are the corresponding French and Spanish terms):
Actually this hierarchy seems to stress the environment and disciplines related to aquaculture.
A different resource is constituted by the so-called reference tables in FIGIS system; the only reference table mentioning aquaculture puts it into another context (taxonomical species):
Biological entity
Taxonomic entity
Major group Order Family Genus Species
Capture species (filter) Aquaculture species (filter) Production species (filter)
Tuna atlas spec
The last resource examined is oneFish directory, which returns the following context (related to economics and planning):
SUBJECT
Aquaculture
Aquaculture development
Aquaculture economics @
Aquaculture planning
With such different interpretations of aquaculture, we can reasonably expect different search and indexing results. Nevertheless, our approach to information integration and ontology building is not that of creating a homogeneous system in the sense of a reduced freedom of interpretation, but in the sense of navigating alternative interpretations, querying alternative systems, and conceiving alternative contexts of use.
To do this, we require a comprehensive set of ontologies that are designed in a way that admits the existence of many possible pathways among concepts under a common conceptual framework. This framework should reuse domain-independent components, be flexible enough, and be focused on the main reasoning schemas for the domain at hand.
Domain-independent, upper ontologies should characterise all the general notions needed to talk about economics, biological species, fish production techniques; for example: parts, agents, attribute, aggregates, activities, plans, devices, species, regions of space or time, etc. While the so-called core ontologies should characterise the main conceptual habits (schemas) that fishery people actually use, namely that certain plans govern certain activities involving certain devices applied to the capturing or production of a certain fish species in certain areas of water regions, etc.
Upper and core ontologies [
Once made clear that different fishery information systems provide different
views on the domain, we directly enter the paradigm of ontology integration, namely
the integration of schemas that are arbitrary logical theories, and hence can have
multiple models (as opposed to database schemas that have only one model) [
In order to benefit from the ontology integration framework, we must transform informal schemas into formal ones. In other words, thesauri and other terminology management resources must be transformed into (formal) ontologies.
To perform this task, we apply the techniques of three methodologies:
OntoClean [
The first one contains principles for building and using upper ontologies for
core and domain ontology analysis, revision, and development. In its current form,
OntoClean also features an axiomatised domain-independent top-level of formal
criteria, concepts and relations (Figure 3) [
ONIONS is a set of methods for enhancing the informal data of terminological resources to the status of formal ontological data types. Some methods are aimed at reusing the structure of hierarchies (e.g., BT/NT relations, subtopic relation, etc.), the additional relations that can be found (e.g., RT relations), and at analysing the compositional structure of terms in order to capture new relations and definitional elements. Other methods concern the management of semantic mismatches between alternative or overlapping ontologies, and the exploitation of systematic polysemy to discover relevant domain conceptual structures.
OnTopic is about creating dependencies between topic hierarchies and ontologies. It contains methods for deriving the elements of an ontology that describe a given topic, and methods to build ’active’topics that are defined according to the dependency of any individual, concept, or relation in an ontology.
In Figure 1, a class diagram is shown of the informal and formal data types taken into account by the forementioned methodologies.
In section 3.1 the types of (meta)data extracted from the resources are described. In the subsequent sections the (meta)data types obtained from the transformation of resources into a merged ontology are also described.
We briefly describe: • the resources that are integrated • how the Integrated Fishery Ontology (IFO) is being built • a mediation architecture to interface the fishery ontology service with the source information systems. 3
The following resources have been singled out from the fishery information systems considered in the project:
the oneFish topic trees (about 1,800 topics), made up of hierarchical topics with brief summaries, identity codes and attached knowledge objects (documents, web sites, various metadata). The hierarchy (average depth: 3) is ordered by (at least) two different relations: subtopic, and intersection between topics, the last being notated with @, similarly to relations found in known subject directories like DMOZ. There is one ’backbone’ tree consisting of five disjoint categories, called worldviews (subjects, ecosystem, geography, species, administration) and one worldview (stakeholder), maintained by the users of the community, containing own topics and topics that are also contained in the first four other categories (Figure 5). Alternative trees contain new ’conjunct’ topics deriving from theintersection of topics belonging to different categories.
OntologicalStructure
Documentation
BT,NT,RT informal axioms
InformalAxioms
(Informal) axioms
as reusable component n n
Reusable component from original n
Fig. 1. A class diagram of the source data types taken into account
1 HAS-PART
Inclusion hierarchies 1
AGROVOC thesaurus (about 500 fishery-related descriptors), with thesaurus relations (narrower term, related term, used for) among descriptors, lexical relations among terms, terminological multilingual equivalents, and glosses (scope notes) for some of them.
ASFA thesaurus, similar to AGROVOC, but with about 10,000 descriptors.
FIGIS reference tables, with 100 to 200 top-level concepts, with a max depth of 4, and about 30,000 ’objects’ (mixed concepts and individuals), relations (specialised for each top category, but scarcely instantiated) and multilingual support. There are modules (water areas, continental areas, biological entities, vessels, commodities, stocks, etc.), also organised by ’views’.
In Figure 2 a diagram is sketched of the methodology used to extract and refine the informal data from the fishery information systems. The methodology is also described in the next sections.
The (meta)data from the resources that have been singled out have been
processed, in order to integrate them within a homogeneous environment, and with a
clear assessment of their nature. In the following we list a set of guidelines that have
been followed to translate and refine data components:
• A detailed evaluation of each source (find the schema -explicit or not- underlying
the implementation of source data, then describe each data type both qualitatively
and quantitatively) is performed.
• A language to represent the KB is chosen that hosts the integration activity. A
description logic like DLR [
Documentation translated entry/ Domain documentation resources defined [al ] do: Translate resources to common format. Trace origin exit/ DOC resources formalised
Lexical sets translated entry/ Lexical resources defined [all] do: Translate resources to common format. Trace origin exit/ Lexicalisation resources formalised
Domain conceived exit/ Resources selected
Resources described entry/ Domain resources col ected do: Use a classification scheme from an ontology of resources exit/ Resources classified Resource processing packages created entry/ Resources classified do: Define activities to be done exit/ Homogeneous resource set defined
Reusable components from resources identified entry/ Homogeneous resource set defined do: Analyse resource schemas exit/ Reusable components identified
Rough list of ontology elements ready entry/ Homogeneous resource set defined, reusable components identified do: Collect all namespaces (concepts,relations,individuals,topics) from resources,
start assigning data types, documentation and terms collected exit/ Rough namespaces created with flags to resources
Core ontologies translated entry/ Core ontology resources defined [FIGIS, top ASFA, top Agrovoc, else] do: Translate core resources to common format exit/ Preliminary core ontology formalised
Domain ontologies translated entry/ Domain ontology resources defined [FIGIS] do: Translate resources to common format exit/ Domain ontology resources formalised
RT axioms translated entry/ Domain RT resources defined [ASFA,Agrovoc] do: Translate resources to common format exit/ RT resources formalised
RT resources refined entry/ RT resources formalised do: Refine with heuristics based on taxonomies and core ontologies exit/ Refined subset of RT axioms ready
Axiomatic resources ready entry/ Domain ontologies translated, RT axioms refined do: Prepare integration space of axioms exit/ Axioms to be integrated
Topic trees translated entry/ Topic resources defined [oneFish] do: Translate resources to common format exit/ Preliminary topic trees formalised
BT/NT hierarchies translated entry/ Domain BT/NT resources defined [ASFA,Agrovoc] do: Translate resources to common format exit/ BT/NT resources formalised
BT/NT hierarchies refined entry/ BT/NT resources formalised do: Refine with heuristics based on core ontologies exit/ Refined subset of BT/NT hierarchies ready
Taxonomical resources ready entry/ Domain ontologies translated, BT/NT hierarchies refined do: Prepare integration space of taxonomies exit/ Taxonomies to be integrated
Topic trees refined entry/ Preliminary topic trees formalised do: Refine trees according to set-theoretic principles exit/ Refined topic trees ready
Assertional resources ready entry/ Domain ontologies translated, BT/NT hierarchies refined, RT resources refined,
DOC and lexicalisation resources formalised do: Prepare integration space of assertions exit/ Assertions to be integrated
List of integratable ontology elements ready entry/ Taxonomical, axiomatic, and assertional resources ready, refined topic trees ready do: Create working namespaces with flags to original resources, maintain links between current resources exit/ Working, interlinked namespaces created with flags to resources
Fig. 2. A diagram of the methodology used to extract and refine the informal data • • •
Extract IS_A taxonomies from AGROVOC and ASFA BT/NT (Broader Term/Narrower Term) hierarchies. Heuristics from upper and core ontologies can be applied to clean up BT/NT hierarchies, for example, the following rule can be applied: if a body part descriptor is NT of an organism descriptor, then this is probably not an IS_A use of NT (probably it is a part-of relation). This results in resources for core and domain taxonomies building.
Expand RT (Related Term) relations from AGROVOC and ASFA. Also non-IS_A BT/NT hierarchies could be refined (expanded) here. Heuristics can be applied here as well, for example, if there exists a systematic relation between to concepts in the core ontology, and there exists a RT relations between two subconcepts of those concepts, then this is an indication for that relation to be the refinement of the RT one. This results in resources for core and domain axioms building.
Reuse UF (Used For) relations and (multi-)linguistic equivalents from all resources. Track must be kept of the context from which a linguistic item has been extracted. This results in resources for ontology lexicalisation.
In the following sections we outline the main steps to build the basic taxonomy, documentation, and architecture for the integrated fishery ontology. 3.3.1 Developing a fishery core ontology (FCO)
In this step, we pick up uppermost concepts and conceptual (categorisation) schemas from sources and integrate them with a certified top-level containing domain-independent concepts, relations and meta-properties. The resources needed for such a task are:
Upper ontology resources: the OntoClean upper level [
Core ontology resources: conceptual templates, (selected in the preliminary phases), relational database schemas, theoretical views on domain topics, domain standards, etc. An informal fishery core ontology (the FIGIS composite concepts) is shown in Figure 4.
In the context of core ontology development, some taxonomical branches (core concepts) have relevant conceptual integration issues that are being studied by ontological engineers and domain experts in close collaboration: • • • biological taxonomies: difficult having a stable framework of reference (in principle, mapping from local taxonomies to a biological one is feasible, but in practice it could be not cost effective) geographic regions: use GIS as a stable framework of reference? geographic names? institutions: maybe automatic clustering of individuals through classification • • • • • • fishing devices (including vessels) fishing and fish farming techniques (plans and activity types) farming systems (sets of components) fishery regulations (norms) fishery managament systems (plans) production centers
Body Ordinary object
Amount of matter Arbitrary collection Physical Object Mental Object Relevant part Place State Process
Accomplishment
Development is performed as incremental loading and classification of upper and core level ontologies in the Ontology Server.
Another indirect resource that can be exploited to build the core ontology is
the analysis of systematic polysemies (they have been already used in the mining of
large medical thesauri, cf.[
This phase deals with the integration of the resources for domain ontology development with the fishery core ontology (developed in the previous phase).
Resulting taxonomies could be either ’tolerated’ or ’cleaned up’. Tolerance amounts to have widespread and unexplained polysemy for terms, but it is not time consuming. Cleaning is the most time consuming task, since a frequent scenario is the following: concept C from source S1 (C^S1) is in principle similar to a D^S2 (usually because they share one or more terms), but they actually occupy two taxonomical places that make them disjoint according to the upper or core ontology.
The ONIONS methodology [
Some cleaning will be needed in any case to remove at least the major taxonomical clashes. This results into a domain taxonomy. Additional effort should be dedicated to distinguish:
Concepts vs individuals (heuristics applicable: country names, institutions, etc.).
Backbone concepts vs viewpoint
contingent notions), cf. [
This eventually results into a refined domain taxonomy.
concepts (roles, reified properties, 3.3.3 Collecting existing documentation and producing glosses
Available resources for ontology documentation are collected and associated as a kind of annotation (gloss) to domain concepts. Concepts lacking a gloss require a new one.
For core concepts and relations, besides existing glosses, an extensive description of their scope in the FCO is provided. 3.3.4 Designing a preliminary topic architecture
A preliminary topology for most general topics (to be used for ontology
modularisation as well) is figured out. Here the following resources are reused:
ontologies for topics (Welty s topic topology [
Administration Subjects
Ecosystem Geography
Species Stakeholders
The topic topology will be used both for maintaining the ontology library and for managing text indexing and retrieval. Figure 5 shows how the current topic spaces of oneFish are structured. Figure 6 shows an ontology-based architecture for the Integrated Fishery Ontology.
Once taxonomies are cleaned to a certain extent, documented, and divided into appropriate namespaces, activities aimed at raising the conceptual detail of the ontology can be started. The most important is the characterisation of domain concepts with axioms. In order to realise this, domain resources containing informal relationships, and (at least some) glosses from documentation are upgraded to the status of logical axioms.
Informal relationships can be found in thesauri (e.g. related term) as well as reference tables and topic trees. They are mined in order to understand: 1) if the axioms are applicable to all the subconcepts of the concept to which the axiom pertain, and 2) what quantification is applicable to those axioms: existential (necessary) or universal (contingent)?
This results into formal Domain Axioms. This axiom set is enhanced by
axiomatising glosses. Here the ONIONS methodology [
This activity is time-consuming, and semi-automatic techniques are still a
research issue [
The axioms obtained from informal relationships and glosses are revised according to the fishery core ontology developed so far.
Following OnTopic methodology [
A dependency chain is the transitive closure of the logical depend-ons of a concept. The transitive closure is applied to the defining elements of a concept. Here a set of relevance parameters are applied in order to
An integrated fishery ontology benefits from the existence of terms already related to concepts in the original resources, since they semi-automatically provide the so-called lexicalisation of concepts. On the other hand, having an integrated ontology also provides a powerful tool to check polysemous senses of terms, as well as to check consistency of UF thesaurus relations and consistency of multi-lingual equivalents.
The basic idea is that user queries, through a query interface, can be submitted to two kinds of servers: if the query aims at retrieving documents, a topicbased fishery agent rewrites the query in order to submit it to heterogeneous databases (brokering); if the query aims at finding specialised conceptual or terminological information, it is directed to the Fishery Ontology Server (FOS). In both cases, the query interface uses FOS. Query rewriting needs also mapping relations from the integrated fishery ontology to the source thesauri.
In this paper we have outlined some research solutions within the framework of ontology integration that are based on formal upper and core ontologies. Some details have been given on how informal schemata such as thesauri, reference tables, and topic trees can be reused and refined in order to be manipulated by ontology integration. Some hints have also been shown about the dependence of topic trees from ontologies, a promising research area for the semantic web.
In fact, the overall research issue underlying the FOS project is to provide a unified methodology of ontology integration and merging based on formal ontologies, ontology library design, topic trees building and maintainance, and efficient web search and indexing.