International Journal of Knowledge Processing Studies

Abstract This study evaluates and compares the performance of two transformer-based multilingual language models: XLM-RoBERTa and MuRIL, for extractive question answering (QA) tasks in Hindi and English. The study focuses on the effectiveness of leveraging pre-trained transformer architectures through task-specific fine-tuning, with particular emphasis on accuracy, language performance, and computational efficiency. The datasets, collected and prepared for bilingual QA experimentation and made available through Kaggle, were divided into training and testing subsets using an 80:20 split. Model performance was evaluated using multiple complementary metrics, including Exact Match (EM), F1 Score, BLEU Score, ROUGE-1, ROUGE-2, and ROUGE-L, along with average response time per question. The experimental results demonstrate that fine-tuning can substantially improve QA performance when task-specific adaptation is beneficial, particularly for MuRIL and Hindi QA, whereas the English XLM-RoBERTa results show that direct evaluation of the pre-trained model can outperform the fine-tuned configurations. However, the experiments also reveal the risk of overfitting when excessive training epochs are employed, as reflected by increasing training loss and deteriorating validation performance. Quantization was additionally investigated as an optimization technique and demonstrated meaningful reductions in inference response time, although its effect on predictive accuracy was model and language dependent. Overall, the findings highlight the potential of multilingual transformer models for developing effective bilingual QA systems and provide practical insights into balancing model accuracy, generalization, inference efficiency, and computational requirements, particularly for Hindi and other relatively low-resource languages.

Abstract This paper explores the intricate mechanisms underlying knowledge transfer in Electronic word-of-mouth (eWOM) campaigns and their far-reaching implications for brand equity in the digital era. The aim is to interpret how eWOM channels facilitate the strategies that can optimize knowledge transfer in eWOM campaigns to the benefit of brand equity. Through an extensive review of the existing literature and expert interview analysis, 187 codes were identified. Using 410 Likert questionnaires, the quantitative method was employed to test the proposed model. Exploratory factor analysis and PCA (Principal Component Analysis) method were used by SPSS software. By AMOS software, first and second-order confirmatory factor analyses were conducted to ensure a consistent factor structure between the items and structures. CR criterion was used to assess reliability. The AVE, GOF, TLI, CFI, NFI, and RFI indices were also used to evaluate the model. Understanding the underlying mechanisms of knowledge transfer in eWOM campaigns is essential for brand managers and marketers seeking to bolster brand equity in the digital landscape. Leveraging these mechanisms effectively can enhance brand reputation, customer loyalty, and overall business success. This research emphasizes the critical importance of eWOM strategies as a key driver of brand equity in the contemporary marketing landscape.

Abstract In the country's policy-making and decision-making arena, many new and dynamic issues are faced by the country's research agents. To solved these problems, must be prepared a cycle between the most theoretical layers and the most operational layers. By designing and formulating the wisdom ecosystem model in the field of decision-making, it is possible to create this operational cycle that prepares the back and forth between these levels. This research was carried out with the objectives of "compilation of the ecological model of wisdom in the field of decision-making of research organizations" with a mixed approach and with an exploratory and contextual method. The two statistical populations of this research are: a) the general research community; 25 people with special characteristics, and b) expert society; There were 15 experts who formed the panel group and tried to produce literature by holding brainstorming sessions and summarizing the mentioned cases, they completed the conceptual model of the research. Based on this research, the dimensions and components of wisdom ecosystem in the field of decision-making of Iranian research organizations are: Senior managers (wise judgment, foresight and insight, rationality, applying experiences, understanding the correlation of affairs, understanding issues), Environment (communication channels, dynamics, external knowledge), Knowledge centers (human resource management, human resources, organizational culture, knowledge management, strategy and leadership, decision-making and policy-making), Actors (universities and research centers of the country, government, industry, Supreme National Security Council, Islamic Council and foreign actors), and upstream documents (laws and regulations, modern Islamic civilization, S&T document and ...).

Abstract Drawing a knowledge map of scientific productivity in the field of knowledge engineering in the 2011-2021 period on Web of Science. This was an applied descriptive-analytical study using a quantitative research design. The research population consisted of all scientific products in the field of knowledge engineering, which included 7724 documents published in indexed journals on the Web of Science database in the 2011-2021 period. The data were analyzed in Excel. VOSviewer was used for constructing the bibliometric networks of researchers, institutions, and countries, Histcite was used for obtaining information through scientometrics methods, and Gephi was used for obtaining betweenness centrality, degree centrality, and eigenvector centrality. Xu, Yang from Beijing University of Science and Technology had the most collaboration with other researchers by publishing 69 documents in the field of knowledge engineering. Two Chinese universities were ranked in first and second place and two Iranian universities were in third and fourth places.
Studies on Knowledge Engineering began in the US, the UK, Japan, France, and Australia in 2010-2012. Meanwhile, publications in this field have been pursued with more intensity by China, Iran, Spain, and Russia since 2014. The findings indicate that many researchers are working in the field of knowledge engineering, with the Chinese researchers being the most active compared to other countries. Meanwhile, Asian countries seem to be more involved in this field. Furthermore, most of the journals of knowledge engineering were conference journals.

Abstract Purpose: This study aims to determine the effect of query expansion on scientific texts retrieval in Persian.
Method: The present study was conducted using a quasi-experimental method. The results are obtained by analyzing 40 initial and expanded queries of postgraduate students in the Faculty of Management, University of Tehran. Query expansion was performed manually using primary research results.
Findings: Query expansion of Persian scientific texts leads to an increase in the number of related retrieved documents, as well as the comprehensiveness and accuracy of retrieving scientific data in Elmnet search engine, which as a result, improves the overall performance of information retrieval.
Results: Nowadays, automatic query expansion is on the agenda of databases. Given that Persian databases are not fully developed, and the existence of specific problems with writing in the Persian language, information literacy training and the method of defining and expressing information requirements and providing them to the information retrieval systems, can have a significant impact on postgraduate students and researchers, to retrieve the required information and save them time and money.

Abstract The use of IoT in intelligent agriculture is now very common among farmers, and with the use of drones and sensors, advanced agriculture is rapidly becoming a growing global standard. The use of IoT requires infrastructure that is defined within an IoT governance framework. This paper proposes new approaches to knowledge management, Artificial Intelligence, and IoT governance and their impact on productivity in the agriculture sector by hiring specialized people who are called AI-aided Agriculture Knowledge Managers.
Given the importance of all three topics, knowledge management, IoT, and agriculture, we have tried to show the impact of the presence of AI-aided Agriculture Knowledge Managers and IoT in reducing water consumption as one of the most important requirements of agriculture in a simulation by Matlab Software. The data was obtained from the Eurostat database. We also provide a framework for the presence of AI-aided Agriculture Knowledge Managers in fields that are specialized in agricultural science and knowledge management. Finally, due to the importance of governance in this sector, a framework for the governance of artificial intelligence in the field of agriculture with the presence of knowledge managers has been proposed.

Abstract Digital content marketing is an example of new communication and online approaches that is expanding and penetrating among its users. This type of marketing is a process to create high-quality and valuable content to guide customers and introduce your brand and attract customers. The current research seeks to identify the knowledge components effective on the antecedents and consequences of effective digital content marketing to facilitate knowledge acquisition. In the present study, by using the systematic review and meta-synthesis approach, the results and findings of previous researchers by performing the 7 steps of the Sandelowski and Barroso method have been analyzed, and the knowledge components that are effective on the antecedents and consequences of effective digital content marketing have been. To measure reliability and quality control, the Kappa method was used, and its value was identified for the indicators at the level of excellent agreement. The results in the Max Qda software led to the identification of 38 knowledge areas containing consequences in the form of seven components and two general dimensions and 21 knowledge areas containing the antecedents of effective digital content marketing in the form of four components and two general dimensions. Finally, the customer's behavioral engagement with the brand was identified as the indicator with the highest frequency among the consequences and attention to the characteristics of the audience as the indicator with the highest frequency among the antecedents of effective content marketing, and the knowledge obtained from them helps the field of digital content marketing.

Abstract The research aims to predict demand and optimize product ordering within the supply chain using artificial intelligence. Employing a purposeful sampling method, 12 managers from Calais were selected as the sample group. This study utilized two Delphi techniques and a neural network. Through semi-structured interviews conducted in a fuzzy Delphi panel, relevant components were identified. Predictions were made using the Multi-Layer Perceptron Neural Network toolbox in MATLAB software. The results from the fuzzy Delphi analysis indicated that the primary factors influencing the forecast included warehouse inventory, sales from the previous week, sales from the previous month, cargo in transit, fluctuations in customer numbers compared to the past, competitors' market status, government regulations, and the company's development plans. After finalizing the Delphi process, the key factors identified were warehouse inventory, sales from the previous week, sales from the previous month, changes in customer numbers compared to the past, competitors' market status, and government regulations. The neural network predictions demonstrated that, due to the fluctuating demand trends, the predicted values closely aligned with the actual values. According to findings, the neural network's predictions were deemed acceptable, even with the rapid fluctuations in actual demand. In the case of Kale Dairy Company, utilizing artificial intelligence can yield significant benefits. Firstly, more accurate demand forecasting through artificial intelligence algorithms can lead to a reduction in waste and excess inventory within the supply chain. This enables the company to better identify and meet customer needs, ultimately resulting in increased customer satisfaction.

Abstract This research aims to develop a data-based model for fifth-generation universities. Creating a data-driven model in a university environment is essential in education. The primary mission of higher education is to address the specific educational needs of individuals, as well as the needs of society and its economic development. The study was conducted in both qualitative and quantitative sections. The grounded theory is conducted based on the perspectives of the chancellors of Islamic Azad University. 21 people were selected using snowball sampling techniques. In the following, a six-category model is provided. Analysis was done using NVIVO software. The statistical population in the quantitative section consisted of all professors from Islamic Azad University nationwide. A sample size of 381 professors was selected using the Cochran sampling formula. The research tool was a questionnaire created by the researcher. Then, using the model presented and the suggested pattern fit, the performance of the model is predicted based on the K-Mean method in Weka and RapidMiner software. According to the results, the proposed model was approved by experts. The analysis of structural equations was also confirmed. According to the Waode algorithm model, the highest accuracy was 81%.