Background:
Keyword search engines are unable to understand the intention of user as
a result they produce enormous results for user to distinguish between relevant and non relevant
answers of user queries. This has led to rise in requirement to study search capabilities of different
search engines. In this research work, experimental evaluation is done based on different metrics to
distinguish different search engines on the basis of type of query that can be handled by them.
Methods:
To check the semantics handling performance, four types of query sets consisting of 20
queries of agriculture domain are chosen. Different query set are single term queries, two term queries,
three term queries and NLP queries. Queries from different query set were submitted to
Google, DuckDuckGo and Bing search engines. Effectiveness of different search engines for different
nature of queries is experimented and evaluated in this research using Grade relevance
measures like Cumulative Gain, Discounted Cumulative Gain, Ideal Discounted Cumulative Gain,
and Normalized Discounted Cumulative Gain in addition to the precision metric.
Results:
Our experimental results demonstrate that for single term query, Google retrieves more
relevant documents and performs better and DuckDuckGo retrieves more relevant documents for
NLP queries.
Conclusion:
Analysis done in this research shows that DuckDuckGo understand human intention
and retrieve more relevant result, through NLP queries as compared to other search engines.