Fatima_2016.pdf (6.63 MB)
An integrated architecture for semantic search
thesis
posted on 2023-08-30, 14:38 authored by Arooj Fatimameaningful manner such that software agents can search, reason with and manipulate this
data based on an understanding of its semantics. Accessing structured data from Linked
Open Datasets currently requires the use of formal query languages (such as SPARQL) which
poses significant difficulties for the end users.
One way to solve this problem is to provide a Natural Language Interface (NLI) to query
semantic data. The author undertook a comprehensive literature survey of existing semantic
search tools and performed a critical analysis to identify their strengths and weaknesses.
Although some of the existing tools support natural language, they are limited in their
techniques for query processing, result ranking, result readability and ease of integration
with other search tools. Based upon this analysis, this research proposes a new architecture
framework called SIRF (Semantic Information Retrieval Framework) for semantic search to
address these shortcomings.
This thesis provides a complete overview of the proposed framework, including: the
research challenges it addresses; its architecture; the techniques to map user queries to
SPARQL queries and to rank domains based on ontology concepts; and the evaluation of the
proposed system through a prototype. Evaluation of the prototype demonstrated the validity
of the approach. However the quality of resulting queries (and consequently retrieved
answers) depended upon the accuracy of the NLP parsers invoked by the prototype. Syntactically
well structured NL queries were correctly parsed, yielding better formed SPARQL
queries. Less structured NL queries performed poorly. As the framework is not tied to any
particular parser, result quality can be improved by utilising better parsers as they become
available.
The author believes that this work can be employed by a variety of end-user applications
that wish to utilise structured data.
History
Institution
Anglia Ruskin UniversityFile version
- Accepted version
Language
- eng
Thesis name
- PhD
Thesis type
- Doctoral
Legacy posted date
2017-02-06Legacy creation date
2017-02-06Legacy Faculty/School/Department
Theses from Anglia Ruskin UniversityUsage metrics
Categories
No categories selectedLicence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC