Explore the vast range of titles available.

Short-text Expansion plays a significant role in enhancing the quality, diversity, and practicality of Short-text, helping users to more comprehensively understand the content expressed in the Short-text. In this paper, we aim to enhance the capabilities of large language models in short-text expansion through knowledge graphs and propose the knowledge-enhanced and prompt-aware large language models. First, we construct a multi-dimensional knowledge graph that includes semantics, sentiment, and topics based on large language models in domain-specific text. Second, we propose a method for mining prompts of Short-text across the three dimensions of semantics, sentiment, and topics based on the constructed multi-dimensional knowledge graph. Finally, we match triplets in the constructed knowledge graph based on the generated prompts in the three dimensions. The matched triplets is then integrated by the large language model to generate a expansion of given short-text. Experiments are conducted using three large language models on two public datasets, and the results indicate that our model shows improvements across multiple metrics for text similarity, readability, and coherence compared to the short-text expansion generated by the baseline large language models and existing methods.

Knowledge graph (KG) is a representative technique of knowledge engineering, and it is often used in various intelligence applications, which assume that all triples in knowledge graphs (KGs) are correct. However, due to the noise brought by automatic KG construction techniques and the fuzziness of knowledge in specific fields, measuring uncertainty of KGs (i.e., the confidence of each triple being true) is important to the tasks of error detection and fact verification. Existing studies on triple confidence measurement either only relies on explicit evidences or merely depends on embedding evidences, which causes the resulting confidences are not precise enough. To solve this problem, in this paper, we propose a new triple confidence measurement (TCM) method, which combines multiple heterogeneous evidences including explicit evidences (i.e., concept paths and neighbor concept subgraphs) and different embedding evidences acquired by large language model, KG embedding models, contrastive learning, and graph convolutional network. Experiments on different real-world datasets demonstrate not only the superiority of TCM in the tasks of error detection and link prediction, but also the effectiveness of all proposed explicit evidences and embedding evidences.

Accurate operation completion time prediction is of significant value for industrial production, improving the plan rationality and production efficiency. Numerous data-driven approaches have been introduced for this task. However, there are a lot of prior knowledge and domain expertise in the production environment, which may greatly facilitate the training progress and improve the model’s performance. How to integrate domain knowledge into the model learning process is a pressing issue. We explore a paradigm for the autonomous integration of data and knowledge in industrial big data, which integrate symbolic business logic and domain knowledge into the data-driven neural model. Based on this paradigm, we propose a method for the operation completion time prediction in automated container terminals. We construct a terminal relational graph, organizing multi-source data into a unified graph structure. The associated features are obtained through traversal searches under the guidance of domain knowledge. A heterogeneous graph neural network is devised to learn data patterns and their underlying symbolic domain knowledge. Experiments conducted on actual terminal industrial data containing over 880,000 records shows the effectiveness of our proposed method compared with other regression models. The results demonstrate that our method can capture associated features and integrates symbolic domain knowledge into the neural model, thereby enhancing the data’s learning ability.

Numerous real-world information networks form Heterogeneous Information Networks (HINs) with diverse objects and relations represented as nodes and edges in heterogeneous graphs. Similarity between nodes quantifies how closely two nodes resemble each other, mainly depending on the similarity of the nodes they are connected to, recursively. Users may be interested in only specific types of connections in the similarity definition, represented as meta-paths, i.e., a sequence of node and edge types. Existing Heterogeneous Graph Neural Network (HGNN)-based similarity search methods may accommodate meta-paths, but require retraining for different meta-paths. Conversely, existing path-based similarity search methods may switch flexibly between meta-paths but often suffer from lower accuracy, as they rely solely on path information. This paper proposes HetFS, a Fast Similarity method for ad-hoc queries with user-given meta-paths on Heterogeneous information networks. HetFS provides similarity results based on path information that satisfies the meta-path restriction, as well as node content. Extensive experiments demonstrate the effectiveness and efficiency of HetFS in addressing ad-hoc queries, outperforming state-of-the-art HGNNs and path-based approaches, and showing strong performance in downstream applications, including link prediction, node classification, and clustering.