Question-Answering for Agricultural Open Data - Transactions on Large-Scale Data-and Knowledge-Centered Systems - page 16

is disadvantageous in that the user must input keywords and iteratively tap and
scroll through a Search Engine Result Page (SERP) to find an answer. There-
fore, this paper proposes a voice-controlled question-answering system for search
of agricultural information. Voice control is suitable for agricultural work since
users typically have dirty hands and can speak freely without disturbing other
people. At the same time, it provides a mechanism for registering the work of
the user, since data logging is the basis of precision farming according to the
Japanese Ministry of Agriculture.
There are many sources of agricultural information on the Web. There are
also many databases (DBs) authorized by agricultural organizations. However,
the sources on the Web are written in natural language and the DBs do not
employ uniform schema, making it impossible to search them using standard-
ized procedures. Moreover, the DBs may have open application programming
interfaces (APIs) for search, but contents are closed in many cases. Therefore,
we propose Linked Open Data (LOD) for agricultural knowledge with a view to
establishing an open standard for agricultural data.
This remainder of this paper is organized as follows. Section 2 introduces
related work regarding standardization of agricultural data and applications sup-
porting the work. In section 3, we propose Plant Cultivation LOD and describe a
mechanism of LOD extraction from the Web. Then, in section 4, we describe the
development and evaluation of the question-answering system for agricultural
work using LOD. Finally, we conclude by referring to future work in section 5.
2 Related Work
A standard for agricultural data, agroXML[ 2 ], which is an XML schema for
describing agricultural work, has been proposed. It is used as a means of exchang-
ing data in a structured and standardized way between farm and external stake-
holders (government, manufacturer, and retailer) in the EU. In practice, only the
necessary part of the schema is exchanged according to the purpose. However,
elements such as “cultivation” and “WorkProcess” in agroXML are mainly for
data logging, and not for the description of cultivation knowledge. agroXML also
adopts a hierarchical XML schema, making it dicult to trace content such as
graphs (although the graph version is under development).
Regarding semantic application of agricultural data, the Food and Agricul-
ture Organization (FAO) of the United Nations[ 3 ] is currently developing the
Agricultural Ontology Service Concept Server, whose purpose is the conversion
of the current AGROVOC thesaurus to Web Ontology Language (OWL) ontolo-
gies. AGROVOC is a vocabulary containing 40000 concepts in 22 languages
covering agricultural subject fields, and expressed in the W3C Simple Knowl-
edge Organization System (SKOS) and also published as LOD. To the best of
our knowledge, however, AGROVOC does not include knowledge of plant culti-
vation.
With regard to filed applications for agriculture, Fujitsu Ltd. offers a record-
ing system that allows the user to simply register work types by buttons on