This work presents a new framework as to how web mining is helpful for information retrieval, using ontology and web log files. Ontology plays a major role in the retrieval of semantic data. The researcher has already constructed the string instrument ontology using protege 5.0, which helps in refining the web search in music domain. The researcher has proposed a novel approach for ontology management in which the ontology is continuously updated using the knowledge extracted/discovered from the analysis of the log file (specifically the data related to the referrer field) in form of new concepts and new relationships between new and/or existing concepts. The goal of this study is to use data mining algorithms to analyse visitors and visited web pages of the website and somehow characterise or distinguish them in some way. During this the researcher has collected ‘guitar’ web access log from guitar selling website of 363 days of the year 2016. After pre-processing of this log file, two new feature sets have been extracted from ‘guitar’ log file and constructed two files namely ‘File1’ and ‘File 2’. File 2 is also known as query log. Further clustering (EM), association rule finding (Apriori) and sequential patterns (n-gram) algorithms have been applied for suggestions of new concepts to continuously update and improve the existing ontology from time to time.