Cluster Extraction and Coding: Automatic grouping of similar sentences
The CLUSTER EXTRACTION tool is a unique retrieval feature that allows one to quickly identify all similar paragraphs or sentences in a large collection of documents. This tool relies on a patent-pending algorithm that can quickly compare up to 32,000 text segments and group the most similar ones into clusters. The amount of items that may be clustered depends on the amount of memory available. If all extracted segments cannot fit into memory, QDA Miner will offer the possibility to select a random sample to be clustered. The grouping performed is resistant to spelling errors and may also bring together various inflected forms of nouns and verbs. Once computed, clusters of items may then be coded or modified using intuitive drag-and-drop operations. Un-clustered items will also be suggested based on their level of similarity to existing clusters.
Such a tool can speed up the coding process of large quantities of open-ended responses, Twitter feeds or short comments by a factor that may vary from two to 50 times faster or more, compared to what someone would achieve by manually coding those responses one at a time.To start the clustering process, select the CLUSTER EXTRACTION command from the ANALYSIS