Designing of Ontology for Domain Vocabulary on Agriculture Activity Ontology (AAO) and a Lesson Learned
This paper proposes Agriculture Activity Ontology (AAO) as a basis of the core vocabulary of agricultural activity. Since concepts of agriculture activities are formed by the various context such as purpose, means, crop, and field, we organize the agriculture activity ontology as a hierarchy of concepts discriminated by various properties such as purpose, means, crop and field. The vocabulary of agricultural activity is then defined as the subset of the ontology. Since the ontology is consistent, extendable, and capable of some inferences thanks to Description Logics, so the vocabulary inherits these features. The vocabulary is also linked to existing vocabularies such as AGROVOC. It is expected to use in the data format in the agricultural IT system. The vocabulary is adopted as the part of “the guideline for agriculture activity names for agriculture IT systems” issued by Ministry of Agriculture, Forestry and Fisheries (MAFF), Japan. Also we investigated the usefulness of the ontology as the method for defining the domain vocabulary.
KeywordsOntology Agriculture Agronomic sciences Knowledge representation Core vocabulary Vocabulary management
The various IT systems have been introduced in farm management to realize better management, i.e., more efficient resource management, finer production control and better product quality. Now data management is indispensable in farm management. Data in farm management is also used in own purpose but the aggregated data is used for statistics, analysis and prediction for area agriculture.
Data in agricultural IT systems is nonetheless not easy to federate and integrate since the languages to describe data are not unified. Terminology in agriculture such as names of activity, equipment, and crop has not been well standardized mainly because agriculture has been local. Some of locality comes from diversity of culture and environment and others from the way of business, i.e., farms are small and run independently. But introduction of IT systems changed the situation; farms can be connected beyond the barrier of individual farms, regions, and even culture. But un-unified terminology exists as the problem. Without unified terminology, smooth data exchange cannot be enabled. So standardization of terminology is the key to enhance agriculture with IT systems. We focus on agriculture activity in this paper. Agriculture activity is the most basic element of farm management and also the most difficult to standardize since it is more abstract than other types of terminology like equipment and crop.
In this paper, we investigate the existing vocabulary system for agricultural activities. Then we propose the agriculture activity ontology by paying attention to the linguistic feature of agricultural activities. We also explore reasoning functions with the ontology and web services to utilize the ontology. Finally, we discuss the future directions of the improvement and extension of the ontology.
2 An Existing Resource: AGROVOC
In this section, we survey the features of AGROVOC (a portmanteau of agriculture and vocabulary)  as an existing agricultural vocabulary system. AGROVOC is the most well-known vocabulary in agriculture supervised by Food and Agriculture Organization (FAO) of the United Nations. AGROVOC is the thesaurus containing more than 32,000 terms of agriculture, fisheries, food, environment and other related fields. It has international interoperability as it is provided in 21 languages. Each term can have the hierarchical structure with narrower concept and broader concept, and there are 25 top-most concepts such as activities, organisms, location, products and so on. 1,434 narrower concepts are provided in activities which contains the concepts about agriculture activity.
In addition, the number of activity names about rice farming, which is important in Asia including Japan, are insufficient. For example, in rice farming, especially pulling seedlings and midseason drainage which are important activity in a rice paddy, are not contained in AGROVOC.
The ambiguousness of relationship among concepts in the existing vocabulary system of agricultural activities is thus problematic. It is required to clearly define the relationships among concepts and specify them. In order to solve these problems, this paper suggests the establishment of the ontology for agriculture activity. The ontology can define the clear concepts by separating concepts and representation. Also it can reflect the characteristics of the domain more clearly by structuralizing the relationships.
3 Designing of Agricultural Activity Ontology
This paper describes the Agriculture Activity Ontology (AAO) as the basis of the core vocabulary of agriculture activity, and it provides semantics for agricultural activity names. Also, AAO is formalized by Description Logics in order to define and classify the agricultural activities clearly. Formalization by Description Logics makes it possible to judge the inconsistency and subsumption among concepts, and to enable more logical inferences. The ontology designed by Description Logics can be converted to OWL so that it can processed by computers.
3.1 The Structuralization of the Agricultural Activities
Our strategy to structuralize agricultural activities is the top-down, i.e., starting from the most general activity and expanding it to more specific activities. The important criteria for the top-down approach is how we can classify more specific concepts consistently. We define more specific concepts by specifying attributes and their values. We furthermore define the general rule for specifying attributes.
We start with the top concept Agriculture Activity which denotes all kind of activities related on crop and/or fields. Then we break down the concept into more concrete concepts. When farmers plan or do a certain agricultural activity, the first decision is what for they would take the action, i.e., purpose is the first attribute to distinguish agriculture activities. After the purpose is well specified, we use other attributes, i.e., act (type of action), target, place, means, equipment, and season in this order. Crop is also introduced so as to define the activity for a specific crop. These eight attributes are used to define the concept and to form the hierarchy of the agricultural activity.
The basic idea of formalization of Agriculture Activity Ontology is that concepts correspond to concepts in Descriptions Logics (DLs) while attributes such as purpose correspond to roles in DLs. By adding the role and the role value, the concepts is defined as the narrower concept of the original concept. If the added role is what is already used in the original concept and the value of the role of the new concept is narrower than that in the original concept, the new concept is also narrower concept of the original concept. It should be noted; not all concepts correspond to terms for farm management since some abstract concepts are introduced just to classify. So we distinguish the abstract concepts not corresponding to terms and concrete concepts corresponding to terms. We call former category and the latter term. For example, thinning and cutting root are terms, and their broader concept activity for uniformity is the category. The category and the term are succeeding the values of the attributes, and they have the relationship of inclusion.
Now let’s look at the ontology in detail. At first, we classify the agriculture activity into two; crop production activity which is related to the crop production directly, and administrative activity which is related to the farm management. crop production activity is classified into the following four activities: crop growth activity which is for the purpose of crop growth, activity for environmental control which is for the purpose of the environment control, activity for post production which is for the purpose of the post production, and activity for support for crop production which is for the purpose of the indirect support in the crop production. So as to define narrower concepts, purpose, act, target, place, means, equipment, season, crop were used as attributes. Classification is conducted by using values of these attributes.
3.2 Polysemic Concepts
4 Reasoning by Agriculture Activity Ontology
5 Web Services Based on Agricultural Activity Ontology
5.1 Namespace of Agriculture Activity
5.2 Version History
Listing of the history of AAO version changes
5.3 Data Sharing
The data of AAO can be downloaded in the RDF/Turtle formats from http://cavoc.org/aao/. This format is well supported by many semantic tools, and it is possible to convert it into other RDF formats if needed. Also, we provide a SPARQL endpoint for users to explore AAO data using SPARQL queries (Fig. 4).
6 Discussion and Future Work
The agriculture activity ontology ver 1.31, the latest version of the agriculture activity ontology, has 355 concepts which are either categories and terms. It covers most of terms in the national agricultural statistics in Japan. Structuralization with attributes are our own idea so that we need discussion and communication with experts in agriculture more extensively to verify the value of the ontology.
The extension to crop-specific ontologies is one of the important directions of AAO. The scope of the agriculture activity ontology is not just the general terms of agriculture but also covers agriculture activities specialized by crop. The activity specialized in the crop currently contains 10 types of crop such as rice, melon, and other things. By using the attribute of the crop, it is possible to extend to the crop-specific ontologies. We are now developing the crop-based ontologies that can define crop-specific concepts by using crop independent concepts.
There are still issues in Structuralization of the ontology. One of them is composite concept. In the agriculture IT systems, there are cases in which the multiple works are managed as a single activity by combining multiple activities. For example, Raising seedling is composite includingSeeding, Fertilization, Watering and other things. However, when the agriculture is planned or implemented, it is managed as Raising seedling. We express this activity by combining existing activities. The concept can be expressed with part-of relationship, but the simple solution is not always suitable since all of the concepts are not sometimes necessary. We are now considering more appropriate formalization for combination of the relevant activities.
International interoperability is next to do. We have already connections with other activities for agricultural ontologies (for example Crop Ontology Group3). Our research has begun from the purpose of establishing the core vocabulary for the field of agriculture of Japan, although it can be independent regardless of language culture so that it can be applied to various languages and cultures. We will improve the ontology in order to develop international core vocabulary.
We designed a domain vocabulary by using ontology based on Description Logics. The design used by the Description Logics can make a concept which has ambiguous meaning classified clearly and deal with the situation when new vocabularies have to be added. The meaning of the concept was defined as suitable attributes for the domain. The structure was constructed to make sense the meaning of the concept by using the value of attributes and it could make the effective processing when the vocabulary lists have to be generated automatically or when the related applications have to be realized. Also it can be used as a tool like dictionary in a namespace for each vocabulary.
We provide the Agriculture Activity Ontology (AAO) so as to standardize the vocabulary for agricultural activities. By using the ontology, it is possible to define concepts of agriculture activities beyond the linguistic diversity of the vocabulary for agricultural activities. The agriculture activity ontology was adopted as the part of “the guideline for agriculture activity names for agriculture IT systems” issued by Ministry of Agriculture, Forestry and Fisheries (MAFF), Japan in 2016, which is one of the achievements of this study . We are now working to extend our idea to other agriculture domains, i.e., the standardization of vocabulary for agriculture such as the crop, distribution, and agricultural pesticide.
Indeed, AAO is basically written in Japanese and expressions of concepts and roles in English are optional. But we here provide the English version of AAO for simplicity of explanation.
SWCLOS is a lisp-based OWL Full processor on top of Common Lisp Object System (CLOS). It is downloadable from https://github.com/SeijiKoide/SWCLOS.
This work was supported by Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP), “Technologies for creating next-generation agriculture, forestry and fisheries” (funding agency: Bio-oriented Technology Research Advancement Institution, NARO).
- 1.AGROVOC Multilingual agricultural thesaurus. Subsequences. J. Mol. Biol. 147, 195–197 (1981). http://aims.fao.org/vest-registry/vocabularies/agrovoc-multilingual-agricultural-thesaurus
- 3.Riesbeck, Christopher K., Roger C. Schank, Inside Case-Based Reasoning, ISBN 0-89856-767-6, LEA (1989)Google Scholar
- 4.OWL Web Ontology Language Guide, W3C Recommendation 10, February 2004 https://www.w3.org/TR/owl-guide/
- 6.Koide, S.: Theory and implementation of object oriented semantic web language Dr.thesis, Department Informatics School of Multidisciplinary Sciences, The Graduate University for Advanced Studies (SOKENDAI) (2010)Google Scholar
- 7.RDF 1.1 Turtle, Terse RDF Triple Language, W3C Recommendation 25, February 2014. https://www.w3.org/TR/turtle/
- 8.Ministry of Agriculture, Fisheries and Forestry (MAFF), Japan: the guideline for agriculture activity names for agriculture IT systems (2016). http://www.kantei.go.jp/jp/singi/it2/senmon_bunka/shiryo/shiryo04.pdf (in japanese)