Abstract Background In order to better help doctors make decision in the clinical setting, research is necessary connect electronic health record (EHR) with biomedical literature. Pseudo Relevance Feedback (PRF) a kind of classical query modification technique that has shown be effective many retrieval models and thus suitable for handling terse language jargons EHR. Previous work introduced set constraints (axioms) traditional PRF model. However, feedback document, importance degree...

Pseudo‐relevance feedback is a well‐studied query expansion technique in which it assumed that the top‐ranked documents an initial set of retrieval results are relevant and terms then extracted from those documents. When selecting terms, most traditional models do not simultaneously consider term frequency co‐occurrence relationships between candidate terms. Intuitively, however, has higher with more likely to be related topic. In this article, we propose kernel co‐occurrence‐based framework...

Abstract Recently, large pretrained language models (PLMs) have led a revolution in the information retrieval community. In most PLMs‐based frameworks, ranking performance broadly depends on model structure and semantic complexity of input text. Sequence‐to‐sequence generative for question answering or text generation proven to be competitive, so we wonder whether these can improve effectiveness by enhancing semantics. This article introduces SE‐BERT, semantically enhanced bidirectional...

Abstract The pre‐trained language model (PLM) based on the Transformer encoder, namely BERT, has achieved state‐of‐the‐art results in field of Information Retrieval. Existing BERT‐based ranking models divide documents into passages and aggregate passage‐level relevance to rank document list. However, these common score aggregation strategies cannot capture important semantic information such as structure have not been extensively studied. In this article, we propose a novel kernel‐based...

Pre-trained models have garnered significant attention in the field of information retrieval, particularly for improving document ranking. Typically, an initial retrieval step using sparse methods such as BM25 is employed to obtain a set pseudo-relevant documents, followed by re-ranking with pre-trained model. However, semantic captured from sentences or passages usually only applied ranking, limited use query expansion. In fact, within documents plays critical role selecting appropriate...

The tamper-proof of web pages schemes mainly includes two aspects, one is the generation algorithm on pre-embedded watermarking, and other program watermarking embedded. In this paper, a novel page watermark scheme proposed for detecting locating tampered web-page. Proposed realize location target HTML tampered. PCA digital ULC (Upper-Lower Coding) embedding are separately executed each column row original source code through comparing calculation results with extracted authentication...

The influence of neutral agents on the evolutionary dynamics social tolerance is discussed based a recently proposed economic interaction model with local cost functions. We show that dynamical structure completely changed even if few are introduced into society. Especially, full intolerance steady state, which stable in previous works, becomes unstable and avoidable due to incentive by agents. necessary condition achieving also reduced compared works without

Recently, the multi-stage reranking framework based on pre-trained language model BERT can significantly improve ranking performance information retrieval tasks. However, most of these BERT-based frameworks independently process query-chunk pairs and ignore cross-passages interaction. The context around each candidate passage is extremely important for relevance judgement. Existing aggregation methods obtain through statistical method lost part semantic information. Therefore, to capture...

Characteristic part family definition, master model feature diagram and XML stored document of are given for product variant design. Then, design process is elaborated, including searching similar parts the similarity retrieval mechanism which studied family. Finally, standard business developed through Java, Pro/E, SQL Server an example on to testify validity characteristic mechanism.

Recently neural information retrieval systems have spurred many successful applications. Retrieval model to obtain a candidate document collection in the first stage, then use BERT sort documents. Generally, sentence score or paragraph obtained using is integrated into get final ranking result. Semantic similarity less often used select query extensions and integrate semantic pseudo-relevance feedback. We propose new strategy this paper, selecting with model. Incorporating weights...

Machine learning and deep are currently widely used to predict data information. The purpose of this paper is present our contribution field by discussing a common event prediction problem-bicycle sharing demand prediction. Bicycle Sharing Project has significantly reduced resource consumption air pollution, but it suffers from the inability responsible authorities for bicycles at each station based on spatial temporal stations, which could result in use unnecessary human resources...