Large language models are rapidly advancing the field of artificial intelligence, with current research focusing primarily on traditional natural language understanding tasks, such as question answering and information extraction. However, in knowledge-intensive domains, such as intangible cultural heritage, hallucination problems due to insufficient domain knowledge persist. To address this, we present ChatTf, a knowledge graph-enhanced intelligent Q&A system, exemplified by Chinese traditional folklore, aimed at reducing factuality hallucinations in this domain. Specifically, we constructed the Traditional Folklore Ontology (TFOnto) and proposed the Zero-shot Traditional Folklore Triplet Extraction (ZFTE) framework. Driven by TFOnto, ZFTE builds a Traditional Folklore Knowledge Graph (TFKG). We then proposed a dual-stage Retrieval-Augmented Generation framework (TFKG-RAG) based on TFKG to provide traditional folklore knowledge to large language models, mitigating factuality hallucinations in folklore Q&A tasks. In the experimental phase, ChatTf achieved an accuracy of 96.7% on a self-built TFCQD test set, outperforming several state-of-the-art baseline methods. This demonstrates the accuracy and reliability of folklore domain question answering.

Remaining useful life (RUL) prediction is of great significance to ensure the safety and reliability of equipment. Graph neural network-based methods show great potential to improve RUL prediction performance by extracting spatiotemporal features from sensor monitoring data. However, current methods construct sensor-based homogeneous graphs without considering equipment component structure data and prior knowledge, which cannot characterize the dependency between sensors and studied equipment accurately. To solve this problem, we propose a temporal knowledge graph (TKG) informer network for RUL prediction. A TKG of equipment health status integrates sensor data with structure data through prior knowledge, so as to characterize various spatiotemporal features accurately. The graph structure and node information (spatial-domain features) of the TKG at each moment is embedded in a low-dimensional temporal graph representation (TGR). An informer network extracts variable information (temporal-domain features) to generate TGR for RUL prediction. The proposed method was evaluated on public datasets and was found to achieve much higher performance than other state-of-the-art models. The TKG datasets are available at IEEE DataPort: https://dx.doi.org/10.21227/jgs2-kt12.

The design or simulation analysis of special equipment products must follow the national standards, and hence it may be necessary to repeatedly consult the contents of the standards in the design process. However, it is difficult for the traditional question answering system based on keyword retrieval to give accurate answers to technical questions. Therefore, we use natural language processing techniques to design a question answering system for the decision-making process in pressure vessel design. To solve the problem of insufficient training data for the technology question answering system, we propose a method to generate questions according to a declarative sentence from several different dimensions so that multiple question-answer pairs can be obtained from a declarative sentence. In addition, we designed an interactive attention model based on a bidirectional long short-term memory (BiLSTM) network to improve the performance of the similarity comparison of two question sentences. Finally, the performance of the question answering system was tested on public and technical domain datasets.