An MPW-DENN and Crop Ontology Used for Efficient Crop Information Retrieval System in the Agricultural Domain

With the rapid development of computer technology and the Internet, agricultural information resources are exploding. So, the retrieval of specific information is a difficult task. The traditional information retrieval (IR) system takes more time to retrieve the information, and retrieval accuracy is not better. To overcome such drawbacks, this paper proposed MPW-DENN and crop ontology for efficient crop IR systems in the agricultural domain. The proposed crop IR system has three phases. First, the crop ontology-based database is created in this data creation phase, and the data deduplication process is done in the collected crop document set using a bloom filter (BF). The second phase is the trained IR system, which consists of five steps. Initially, in preprocessing, the noise is removed, and then features are extracted. After that, based on the extracted features, the tree, hash code, and weight (THW) is generated. In that THW, the hash code is generated using the message digest 5 (MD5) algorithm. After that, the document is clustered using the matrix objective form of fuzzy C-means (MOFFCM) algorithm, and then the document is trained using the MPW-DENN algorithm. Third, the similarity between the trained hash code of the document and the user query hash code is calculated. Based on this similarity calculation, the crop information is retrieved. Experimental results prove that the proposed system achieved better performance than the state-of-art methods.