Knowledge creation in vocational education using multi-source data fusion under big data environment

The development of big data technology has brought new challenges to the construction of vocational education (VE), knowledge graph (KG), and single data source does not fully capture data characterisation information. Therefore, this paper crawls internet text data from multiple sources VEs. The improved BiLSTM-CRF model is utilised to recognise the entities, and the context-aware location information is introduced into the BERT model to obtain the entity vectors containing context-aware semantics. The similarity function is used to realise entity alignment, TEXTCNN is used to extract semantic features of entity context-awareness, and the entity embedding vectors are obtained through graph annotation network, and the two are fused to obtain a more accurate representation of entity embedding vectors. The experimental results show that the entity recognition accuracy of the proposed method is improved by 3.47%-12.56%, and more accurate VEKG can be constructed.