Knowledge Discovery and Data Mining is one of the recent emerging fields for mining useful information and knowledge from the digital data, which have rapid growth. Even many techniques are available in data mining for the retrieval record based on text pattern mining, still updating discovered patterns is a trouble in datamining. In order to tackle the problems, a record retrieval method using clustering based on text pattern mining is proposed with four stages. In addition with this, this proposed method is performed on the records, based on the two main phases, which are training and testing phases. In the training phase, closed item sets from each record is extracted using support values, followed by the identification of Normalized D-Patterns of records. Noise Negative records are also consider in the proposed work in order to accomplish updated item sets for each records, which leads to the reduction in error making items. After getting accurate updated records, the weight for every record are computed to move for further processes. And then in testing phase, the records which have similar or nearly similar weights are clustered using K-Means clustering. Each record is assigned with a rank, based on the weight of records and the centroid value of clusters in KMeans. As a result, top ranked records are retrieved from the testing phase. Our proposed work is implemented in Java Platform over real-world datasets and the performance of our work is evaluated using the performance measures such as Precision, Recall and F-measure, which shows that our proposed TPMC approach is best for the record retrievals. © Research India Publications.