In digital library a number of authors may have same names. The authors with same name may belong to different domains. It leads to name ambiguity while searching the books in digital library. To formalize these problems a unified probabilistic framework is proposed. Using this framework the books titles are analyzed to find out the similarity as well as strong relationship between them. The books titles with strong relationship and higher similarity are grouped to the same cluster. This process continues until all the books in the library are clustered. The users can get the results based on the domain name search in addition with author name search. The users obtain the results by specifying author name and the domain. For this purpose the data in the digital library are also partitioned according to the domain of authors. And hence the authors with the same name are determined easily