Retrieval Functions Research Articles

A guideline-informed language model for paediatric cardiology demonstrates high performance in answering complex medical questions

Abstract Background Paediatric cardiology presents unique challenges with its diverse and complex cases, limited evidence base, and the necessity for multi-expert involvement in decision-making processes. In this context, the introduction of generative pre-trained transformer (GPT) based large language models (LLMs) offers a potential avenue for the provision of complex information and clinical decision support. Purpose This study evaluates the quality of three different GPT LLMs in answering complex medical questions, including a state-of-the-art preview model that incorporates the German paediatric cardiology guidelines. Methods Seven paediatric cardiologists and paediatric cardiac surgeons generated 72 questions, including complex questions and medical cases with associated questions. The questions were categorized by difficulty and required knowledge (factual and experience-based or mostly experience-based). We prompted the questions to three LLMs: GPT 3.5, GPT 4 and a GPT 4 turbo preview. The GPT 4 turbo preview was customized by incorporating all guidelines from the German Society for Paediatric Cardiology by a retrieval function. Employing one complex instruction for all questions, we prompted the LLMs to provide precise and detailed expert-level responses. The responses from each model were evaluated by experts based on relevance, factual accuracy, severity of possible harm, completeness, superfluous content, and age-related appropriateness from 0 (very bad) to 7 (very good). Differences were calculated using the Kruskal-Wallis-test in SPSS Version 28. Results Our findings indicated a good performance of all models regarding the dimensions tested. The figures show the average ratings (Figure 1, Figure 2A) and highlight significant differences after Bonferroni correction in bold (Figure 2B). The GPT 4 turbo preview, including the retrieval of guidelines, provided significantly more relevant (average rating [AR] 5.94, meaning mostly relevant), accurate (AR 5.6, meaning between somewhat and mostly accurate) and complete (AR 5, meaning fairly complete) answers compared to GPT 3.5 and GPT 4. In terms of difficulty levels or the type of questions, there was no significant difference in rating. Relevance ratings were slightly better in factual questions (AR 5.7) than in those requiring more experience-based knowledge (AR 5,3). Although GPT4 had higher average scores compared to GPT 3.5 in all dimensions except superfluous content, the differences in rating were not statistically significant. All models had relevant difficulties considering the age-related aspects of the questions (AR 4.06-4.45, p=0.455). Conclusion This study highlights the potential and limitations of AI language models in addressing complex medical questions in fields characterized by highly individualized decision-making scenarios. The findings advocate for the development of more specialized AI tools in medicine, tailored to specific medical fields and patient age groups.Fig 1:Average ratings of LLMsFig 2:Rating differences between LLMs

Memristive Circuit Design for Personalized Emotion Generation With Memory and Retrieval Functions

Memristive Circuit Design for Personalized Emotion Generation With Memory and Retrieval Functions

Application of knowledge graph in smart irrigation district management decision making

Regarding the traditional irrigation and drainage management methods in China's irrigation districts, there are problems such as complex management processes, dispersed information systems, difficulty in resource sharing, diverse data sources, and insufficient intelligent auxiliary decision-making. In order to further realize refined intelligent management decision-making on issues related to smart irrigation districts, this article uses the basic information of 28 Yellow River diversion irrigation districts in Henan Province, as well as the overall plan reports and related issue management manuals of 7 management offices of the first and second phase projects of Zhaokou Irrigation District as data sources. Collect and organize data from irrigation areas, and use these to build a knowledge graph for management of issues in irrigation areas diverted from the Yellow River. At the same time, the BERT + BiLSTM + CRF model is used to intelligently identify entities such as irrigation projects and problem events in the Zhaokou Irrigation District inspection log text, and entity alignment technology is used to match the inspection text with entities in the knowledge graph. Finally, combined with the graph retrieval function, the intelligent generation of smart irrigation district management decision-making solutions is realized. The validation of specific irrigation district examples and analysis of the evaluation indexes of relevant models demonstrate the reliability of the problem management decision-making scheme proposed in this paper. This application effectively facilitates the integration of information data in the Yellow River diversion irrigation area and enables visual management of intelligent irrigation zones, presenting a novel concept for the informationization construction of China's irrigation areas.

AMGPT: A large language model for contextual querying in additive manufacturing

Generalized large language models (LLMs) such as GPT-4 may not provide specific answers to queries formulated by materials science researchers. These models may produce a high-level outline but lack the capacity to return detailed instructions on manufacturing and material properties of novel alloys. We introduce “AMGPT”, a specialized LLM text generator designed for metal AM queries. The goal of AMGPT is to assist researchers and users in navigating a curated corpus of literature. Instead of training from scratch, we employ a pre-trained Llama2-7B model from Hugging Face in a Retrieval-Augmented Generation (RAG) setup, utilizing it to dynamically incorporate information from ∼50 AM papers and textbooks in PDF format. Mathpix is used to convert these PDF documents into TeX format, facilitating their integration into the RAG pipeline managed by LlamaIndex. A query retrieval function has also been added, enabling the system to fetch relevant literature from Elsevier journals based on the context of the query. Expert evaluations of this project highlight that specific embeddings from the RAG setup accelerate response times and maintain coherence in the generated text.

Examining the merits of feature-specific similarity functions in the news domain using human judgments

Online news article recommendations are typically of the ‘more like this’ type, generated by similarity functions. Across three studies, we examined the representativeness of different similarity functions for news item retrieval, by comparing them to human judgments of similarity. In Study 1 (N=401\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$N=401$$\\end{document}), participants assessed the overall similarity of ten randomly paired news articles on politics and compared their judgments to different feature-specific similarity functions (e.g., based on body text or images). In Study 2, we checked for domain differences in a mixed-methods survey (N=45\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$N=45$$\\end{document}), surfacing evidence that the effectiveness of similarity functions differs across different news categories (‘Recent Events’, ‘Sport’). In Study 3 (N=173\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$N=173$$\\end{document}), we improved the design of Study 1, by controlling for how news articles were matched, differentiating between dissimilar news articles and articles that were matched on a shared topic, named entities, and/or date of publication, across ‘Recent Events’ and ‘Sport’ categories. Across all studies, we found that users mostly used text-based features (e.g., body text, title) for their similarity judgments, while BodyText:TF-IDF was found to be the most representative for their judgments. Moreover, the strength of similarity judgments by humans and similarity scores by feature-specific functions was strongly affected by how news article pairs were matched. We show that humans and similarity functions are better aligned when two news articles are more alike, such as in a news recommendation scenario.

Application of Deep Learning and Computer Data Mining Technology in Electronic Information Engineering Management

This article studies the application of deep learning and computer data mining technology in electronic information engineering management to meet the library’s demand for larger collection space and alleviate the management pressure of book preservation, borrowing, and return. This article also utilizes the general information mining function to further improve the information retrieval function. The conclusion of this article is as follows: Based on the traditional April algorithm, an improved address based April algorithm is proposed. The improved Apriori algorithm can reduce the final number of permanent data packets and save about 70% of time. For the minimum supported changes, the improved Apriori algorithm has lower execution time and approximately 60% time reduction. This article develops a data mining algorithm that is more suitable for electronic library management information systems based on existing data mining technologies. Has clear theoretical and practical significance.

Development of X-Band Geophysical Model Function for Sea Surface Wind Speed Retrieval with ASNARO-2

In the present study, a new geophysical model function (GMF) is developed for the X-band synthetic aperture radar (SAR) on board the Advanced Satellite with New System Architecture for Observation-2 (ASNARO-2) to retrieve accurate offshore wind speeds. Equivalent neutral wind speeds based on the local forecast model (LFM) are employed as reference wind vectors, and 12,259 matching points from 502 SAR images obtained with horizontal transmitting, horizontal receiving polarization around Japan are collected. To ensure convergence of the calculation, 8129 points are selected from the matching points to determine the basic formula for the GMF and 23 coefficients based on the relationships among the normalized radar cross section, wind speed, incidence angle, and relative wind direction. Compared with the reference wind speeds, the GMF wind speeds showed reproducibility with a bias of −0.10 m/s and an RMSD of 1.37 m/s. Additionally, it can be confirmed that the retrieved wind speed has the bias of 0.03 and the RMSD of 1.68 m/s when compared to the in situ wind speed from the Kuroshio Extension Observatory (KEO) buoy. The accuracy of these retrieved wind speeds is comparable to previous studies, and it is indicated that the developed GMF can be used to retrieve offshore winds from ASNARO-2 images.

Can LLMs Answer Investment Banking Questions? Using Domain-Tuned Functions to Improve LLM Performance on Knowledge-Intensive Analytical Tasks

Large Language Models (LLMs) can increase the productivity of general-purpose knowledge work, but accuracy is a concern, especially in professional settings requiring domain-specific knowledge and reasoning. To evaluate the suitability of LLMs for such work, we developed a benchmark of 16 analytical tasks representative of the investment banking industry. We evaluated LLM performance without special prompting, with relevant information provided in the prompt, and as part of a system giving the LLM access to domain-tuned functions for information retrieval and planning. Without access to functions, state-of-the-art LLMs performed poorly, completing two or fewer tasks correctly. Access to appropriate domain-tuned functions yielded dramatically better results, although performance was highly sensitive to the design of the functions and the structure of the information they returned. The most effective designs yielded correct answers on 12 out of 16 tasks. Our results suggest that domain-specific functions and information structures, by empowering LLMs with relevant domain knowledge and enabling them to reason in domain-appropriate ways, may be a powerful means of adapting LLMs for use in demanding professional settings.

Green Supplier Selection for Small Enterprises without Impairing Earnings

In recent years, more and more studies have shown that sustainable development and competitive profitability don't need to be mutually exclusive, and as a result, not only governments and non-profit organizations, but also more and more companies have started to look at green procurement as an important part of sustainable development. However, as a major contributor to carbon emissions and a key implementer of green procurement, MSMEs have long faced internal and external disadvantages in the form of a lack of expert support and external information. To prevent green procurement from becoming a strategy of information monopoly for large enterprises, and to reduce the cost of green procurement and encourage more SME to join, this paper focuses on the core link of green procurement - supplier selection, and uses AI model to enhance the traditional AHP hierarchical analysis method to provide a set of convenient and cost-effective green supplier selection solutions for SME. high performance green supplier selection program, and provides some observations and suggestions based on the model. In the information age, for individuals or organizations that lack information, large language models such as GPT provide a fast track. The nature of the decision-making capability of large language models, such as machine learning, is increasingly proving to be effective in addition to the traditional function of information retrieval. In this paper, we used ChatGPT to generate a group of virtual experts as an AHP expert panel, and established the method of developing AHP top guides and consistency tests to ensure that their scores are effective. Finally, we determined the 5 first-level indicators and 16 sub-indicators to be considered by micro and small enterprises to make a green supplier selection with guaranteed benefits and no cost increase, as well as the weights of these indicators. The method is highly feasible and scalable, and micro and small enterprises can easily reproduce the programme in practice and add to the learning materials or objectives according to the actual situation or to the specific needs of the enterprise.

A modified neural circuit framework for semantic memory retrieval with implications for circuit modulation to treat verbal retrieval deficits.

Word finding difficulty is a frequent complaint in older age and disease states, but treatment options are lacking for such verbal retrieval deficits. Better understanding of the neurophysiological and neuroanatomical basis of verbal retrieval function may inform effective interventions. In this article, we review the current evidence of a neural retrieval circuit central to verbal production, including words and semantic memory, that involves the pre-supplementary motor area (pre-SMA), striatum (particularly caudate nucleus), and thalamus. We aim to offer a modified neural circuit framework expanded upon a memory retrieval model proposed in 2013 by Hart etal., as evidence from electrophysiological, functional brain imaging, and noninvasive electrical brain stimulation studies have provided additional pieces of information that converge on a shared neural circuit for retrieval of memory and words. We propose that both the left inferior frontal gyrus and fronto-polar regions should be included in the expanded circuit. All these regions have their respective functional roles during verbal retrieval, such as selection and inhibition during search, initiation and termination of search, maintenance of co-activation across cortical regions, as well as final activation of the retrieved information. We will also highlight the structural connectivity from and to the pre-SMA (e.g., frontal aslant tract and fronto-striatal tract) that facilitates communication between the regions within this circuit. Finally, we will discuss how this circuit and its correlated activity may be affected by disease states and how this circuit may serve as a novel target engagement for neuromodulatory treatment of verbal retrieval deficits.

Multimodal bird information retrieval system

Multimodal bird information retrieval system can help people popularize bird knowledge and help bird conservation. In this paper, we use the self-built bird dataset, the ViT-B/32 model in CLIP model as the training model, python as the development language, and PyQT5 to complete the interface development. The system mainly realizes the uploading and displaying of bird pictures, the multimodal retrieval function of bird information, and the introduction of related bird information. The results of the trial run show that the system can accomplish the multimodal retrieval of bird information, retrieve the species of birds and other related information through the pictures uploaded by the user, or retrieve the most similar bird information through the text content described by the user.

The birth of zinc anode-based electrochromic devices

Since the discovery of electrochemical coloration phenomenon, electrochromic devices capable of monitoring transmittance, reflectance, and absorption at designated wavelengths have embraced great achievements. The marriage of electrochemistry and optical modulation has infused fascinating properties in electrochromic devices, which find applications in thermal management, display, smart windows, and camouflage. Inspired by the multipronged advancements in electrochemical devices, the incorporation of multivalent metal ions having rich electrochemistry into electrochromic devices is bloomed in recent years. Zinc, distinguished by its high crustal abundance, suitable standard redox potential, and inherent safety, has facilitated the assembly of highly efficient electrochromic devices. Zinc anode-based electrochromic devices with dual-band (visible and near-infrared) tunability, energy retrieval functions, multi-color options, multiple working modes (transmittance mode and reflectance mode), and scalability have been prominently showcased. Here in this review, the birth of zinc anode-based electrochromic devices will be systematically narrated, starting from the discovery of electrochromic phenomenon, to the evolution of electrochromic devices, and to the latest achievements in zinc anode-based electrochromic devices. Additionally, this review delves into the future development trends and perspectives of zinc anode-based electrochromic devices. This review serves as a handbook, which summarizes the history of electrochromism, introduces the physics behind it, highlights the development in zinc anode-based electrochromic devices, and aims to inspire future endeavors into this field, particularly those focused on developing energy-efficient electrochromic devices.

Research on the Construction and Effectiveness of Teaching Resource Sharing Platform for Art and Design Majors

Abstract In the information society, the sharing characteristics of Internet information provide students with rich teaching resources. This study integrates Hadoop technology with the teaching resource sharing system for art and design majors in colleges and universities. It builds an open-source parallelised computing and distributed architecture based on the traditional B/S model, using the distributed file system HDFS to store massive teaching resources. On this basis, MapReduce, a parallelised programming model, is used to provide parallel computing retrieval functions, and the innovative contract technology of blockchain is introduced to store data-heavy shared teaching resources in the underlying IPFS to ensure the security and trustworthiness of shared data. The results show that under the blockchain environment, the retrieval time of teaching resources is as low as 12ms only, and the Gas value consumed by the operation of storing data totals 0.1385 ETH, which meets the performance requirements of the platform and provides a new practical solution for sharing teaching resources.

Research on the Relationship between the Participation Degree and Experience of China’s Intangible cultural heritage Tourism Activation in the Information Age

Abstract With the development of the trend of cultural and tourism integration, intangible cultural heritage has become an important part of the development of experiential tourism. This paper combines information visualization technology to construct a framework for the design of non-heritage information visualization. By constructing a relevant spatio-temporal retrieval function to reduce the redundancy of non-heritage spatial data storage, combined with the base state correction model, the change process of non-heritage spatial features is retraced. GRU is applied as a unit of temporal feature correlation analysis to model the serial data of the degree of participation in non-heritage tourism revitalization. Two gating controls, the reset gate and the update gate, are used to regulate the flow of NRM information. The relationship between participation level in NRH tourism revitalization and experience was focused on in the information visualization design. The results show that the p-value of tourism experience and tourism revitalization is 0.521**, which is the most significant among all p-values, indicating that tourism experience has the closest relationship with tourism revitalization, and there is a significant positive correlation between both. Among the different groups of tourists, most of the experience scores of group C tourists are concentrated in the range of (80,90), while the advanced level of participation brings visual enjoyment to tourists while conveying the cultural spirit and emotional experience.

Developing a Linked Open Data Platform for Folktales in the Greater Mekong Subregion

This research paper presents the development of a linked open data (LOD) platform that aims to organize and facilitate access to valuable knowledge about folktales and ethnic groups in the Greater Mekong Subregion countries. The study’s methodology involved the creation of a linked open data platform, structuring folktales’ knowledge, and evaluating its performance through expert assessment. The LOD platform was constructed through Google OpenRefine to establish connections with external data sources, and the RDF files (N-Triples) were deployed on Fuseki Server (Apache Jena) to serve as the SPARQL endpoint for querying the linked open data. The Pubby web app was chosen for further development to provide a user-friendly interface, which customized with the Bootstrap framework, featuring an intuitive homepage and a search box function for simplified data retrieval. For the expert evaluation, the study confirmed that the platform performs a high suitability in terms of congruence, reliability, integrity, understandability, collaboration, accessibility, and connectedness. The developed LOD platform exhibits significant potential for expanding its application to various content domains, offering a valuable resource for accessing and exploring the rich cultural heritage of folktales in the Greater Mekong Subregion countries. Doi: 10.28991/ESJ-2023-07-06-06 Full Text: PDF

Verification of C-Band Geophysical Model Function for Wind Speed Retrieval in the Open Ocean and Inland Water Conditions

Verification of the C-band geophysical model functions (GMF) for the open ocean and the Gorky reservoir was carried out using Sentinel-1 IW-mode Synthetic aperture radar (SAR) images. CMOD5.N, CMOD7, GMF for the Caspian Sea, and CSARMOD2 were considered. The motivation for this study is concerned with the clarification of applying C-band GMFs for SAR images including for the conditions of inland water bodies, as well as with the study of the influence of various wind speed direction sources on the results of wind speed magnitude retrieval for ocean conditions. Comparison of wind speed from the CMOD5.N algorithm using wind direction data from NOAA NDBC oceanographic buoys together with the data provided by NCEP reanalysis data showed that regardless of the geographic location, the result does not depend significantly on the choice of the wind direction source. Novel results of CMOD5.N, CMOD7, GMF for the Caspian Sea, and CSARMOD2 applications to the conditions of the Gorky reservoir are presented. The comparison of these results with the meteorological station measurements showed the best agreement for CMOD5.N. The preliminary results on the construction of new C-band GMF adjusted to the Gorky Reservoir have shown statistical parameters better than for Caspian Sea GMF and CSARMOD2.

StopGuess: A framework for public-key authenticated encryption with keyword search

Public key encryption with keyword search (PEKS) allows users to search on encrypted data without leaking the keyword information from the ciphertexts. But it does not preserve keyword privacy within the trapdoors, because an adversary (e.g., untrusted server) might launch inside keyword-guessing attacks (IKGA) to guess keywords from the trapdoors. In recent years, public key authenticated encryption with keyword search (PAEKS) has become a promising primitive to counter the IKGA. However, existing PAEKS schemes focus on the concrete construction of PAEKS, making them unable to support modular construction, intuitive proof, or flexible extension. In this paper, our proposal called “StopGuess” is the first elegant framework to achieve the above-mentioned features. StopGuess provides a general solution to eliminate IKGA, and we can construct a bundle of PAEKS schemes from different cryptographic assumptions under the framework. To show its feasibility, we present two generic constructions of PAEKS and their (pairing-based and lattice-based) instantiations in a significantly simpler and more modular manner. Besides, without additional costs, we extend PAEKS to achieve anonymity which preserves the identity of users; we integrate it with symmetric encryption to support data retrieval functionality which makes it practical in resource-constrained applications.

Awareness and Usage of Information Retrieval Functionalities Found in the Online Public Access Catalogue by Undergraduates at the University of Venda

Calls for a shift from the Online Public Access Catalogue (OPAC) to discovery systems craft a need for recognising the information-searching behaviour of the current genre of library users. This study examined the undergraduate students’ awareness and usage of information search functionalities that are embedded in OPAC at the University of Venda (UNIVEN) in Limpopo Province, South Africa (SA). Quantitative data were collected using a Google Form questionnaire, the link of which was e-mailed to the participants to complete. A systematic random sampling method was applied to arrive at 563 respondents. The findings revealed that most respondents rated their level of awareness of OPAC as average. Although some were familiar with OPAC search functionalities, such as author-and-title search entries, and preferred using simple search options on OPAC, there were respondents who were not familiar with Boolean operators, truncation and wildcard symbols, proximity operators, and field search. The major variables affecting their information retrieval on OPAC were poor internet connection and their lack of skills in searching for information from OPAC. This study recommends that librarians promote the importance of using all OPAC functionalities and train undergraduates to improve their awareness and optimal usage of OPAC.

A Privacy-Preserving Image Retrieval Scheme Based on 16×16 DCT and Deep Learning

In recent years, people tend to upload images to cloud servers, which provide storage and retrieval functions. To prevent users' privacy from leaking to the server, research on cipher-image retrieval has attracted much attention. This work presents a novel encrypted image retrieval method. With this scheme, we perform encryption during the JPEG compression process by applying 16×16 DCT (Discrete Cosine Transform) for blocks' transformation, followed by coefficients distribution and 8×8 blocks' permutation. For the retrieval part, when an encrypted query image is sent by an authorized user, the server extracts its DCT histograms as features and inputs them into our trained network model, which incorporates transpose Multilayer perceptron modules ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$Transpose$</tex-math></inline-formula> <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$MLP$</tex-math></inline-formula> ), for retrieval. Experimental results show that our scheme, compared with related schemes, can improve the retrieval performance significantly, when ensuring compression friendliness and no feature information leakage. Moreover, our scheme enables cipher-image retrieval from multiple image owners.

One for more: Structured Multi-Modal Hashing for multiple multimedia retrieval tasks

Hashing has been widely studied in support of efficient multimedia retrieval due to its high computation and storage efficiency. However, existing multimedia hashing models are designed only for specific multimedia retrieval tasks, e.g., cross-modal retrieval and composite multi-modal retrieval. To simultaneously support multiple multimedia retrieval tasks, multiple different hashing models should be equipped in a multimedia retrieval system. This paper proposes a Structured Multi-Modal Hashing (SMMH) method, which can learn only a hash model for multiple multimedia retrieval tasks. Specifically, we generate structured hash codes by independently projecting the modality data into the Hamming space, where each part of hash codes preserves the specific characteristic of the specific modality data, and they are collaborated to represent the whole multi-modal data. To bridge the heterogeneous gap between multiple modalities, a fusion multi-modal space is constructed based on structured hash codes with label and pair-wise semantic supervision. Besides, we learn hash functions for composite multi-modal retrieval by projecting heterogeneous modalities into the collaborative multi-modal features. Finally, we develop a discrete hash optimization strategy to efficiently solve binary hash codes. Extensive experiments show that our model outperforms the strong baseline methods on multiple multimedia retrieval tasks. Source codes are available at https://github.com/ChaoqunZheng/SMMH.