Semantic relation based personalized ranking approach for engineering document retrieval

Hahm, Gyeong June; Lee, Jae Hyun; Suh, Hyo Won

doi:10.1016/j.aei.2015.01.003

Cited by 15 publications

(6 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Engineering processes involve significant technical and organizational knowledge and comprise a sequence of activities, such as design, analysis, and manufacturing [1]. During these engineering activities, a large amount of data and knowledge is generated and stored in various types of technical documents, such as technical reports, emails, papers, and patents [2]. Prior studies have reported that engineers spend two-thirds of their time communicating to obtain a related document input for their work and make decisions based on such materials [3].…”

Section: Introductionmentioning

confidence: 99%

Deep Learning for Technical Document Classification

Jiang

Luo

et al. 2024

IEEE Trans. Eng. Manage.

View full text Add to dashboard Cite

In large technology companies, the requirements for managing and organizing technical documents created by engineers and managers in supporting relevant decision making have increased dramatically in recent years, which has led to a higher demand for more scalable, accurate, and automated document classification. Prior studies have primarily focused on processing text for classification and small-scale databases. This paper describes a novel multimodal deep learning architecture, called TechDoc, for technical document classification, which utilizes both natural language and descriptive images to train hierarchical classifiers. The architecture synthesizes convolutional neural networks and recurrent neural networks through an integrated training process. We applied the architecture to a large multimodal technical document database and trained the model for classifying documents based on the hierarchical International Patent Classification system. Our results show that the trained neural network presents a greater classification accuracy than those using a single modality and several earlier text classification methods. The trained model can potentially be scaled to millions of real-world technical documents with both text and figures, which is useful for data and knowledge management in large technology companies and organizations.

show abstract

Section: Introductionmentioning

confidence: 99%

Deep Learning for Technical Document Classification

Jiang

Luo

et al. 2024

IEEE Trans. Eng. Manage.

View full text Add to dashboard Cite

show abstract

“…Knowledge retrieval and knowledge recommendation are two main approaches to knowledge reuse [3]. The advantage of knowledge retrieval is that it fully meets the user's retrieval needs and can quickly and accurately find the best matching results from the retrieval keywords [4], [5]. However, the…”

Section: Introductionmentioning

confidence: 99%

A Correlation-Experience-Demand Based Personalized Knowledge Recommendation Approach

et al. 2019

View full text Add to dashboard Cite

Knowledge recommendation is an important means of knowledge reuse that can improve the efficiency and quality of product design. However, at present, there is no good way to fully consider the personalized demands of designers while ensuring the applicability of the recommendation results. Previous studies have usually been based on the similarity between tasks and knowledge or use collaborative filtering technology to accomplish knowledge recommendation. However, these methods do not consider the personal experience of designers and the characteristics of knowledge. This paper proposes a knowledge recommendation approach that integrates the degree of correlation between knowledge and tasks, the feedback-based personal experience, the collective experience of designers, and the degree of demand for knowledge based on the forgetting curve. A knowledge assistance score is generated based on these factors, and the knowledge recommendation list is obtained by ranking the knowledge in descending order of this score. Finally, the approach is applied to a machine shop layout design task and a computer numerical control (CNC) machine tool's spindle design and bearings selection task. The experimental results on two tasks demonstrate that the proposed approach outperforms three baselines on three ranking oriented evaluation metrics. This approach can effectively shorten the time for designers to acquire knowledge by recommending applicable knowledge to assist designers in completing design tasks with high quality and efficiency. INDEX TERMS Collaborative filtering, degree of assistance, degree of correlation, degree of demand, knowledge recommendation, ontology model.

show abstract

“…These files may be of types of reports, CAD files, presentations, scan copies, spreadsheets, art works, images, audios, videos and so on. It has been reported that stakeholders (e.g., engineers, managers, administrators) spend around 30% of time in file search [10,11,12]. However, searching through a large data infrastructure is a time consuming operation, as (i) search queries often return a long list of files, (ii) the list usually includes false positives, and (iii) the queries often miss some target files (false negatives).…”

Section: Introductionmentioning

confidence: 99%

“…Previous research have been focused on different mechanisms to improve the search time by enhancing the retrieval capability of enterprise search engines for engineering document [11,12], multi-topic documents [13], patents [14], mechatronic data [15], BIM document [16,17,18] and so on. However, existing search engines typically do not provide information about historical search activities.…”

Section: Introductionmentioning

confidence: 99%

Ontology-assisted provenance visualization for supporting enterprise search of engineering and business files

Khan

Kanturska

Waters

et al. 2016

Advanced Engineering Informatics

View full text Add to dashboard Cite

In many large engineering enterprises, searching for files is a high-volume routine activity. Visualization-assisted search facilities can significantly reduce the cost of such activities. In this paper, we introduce the concept of Search Provenance Graph (SPG), and present a technique for mapping out the search results and externalizing the provenance of a search process. This enables users to be aware of collaborative search activities within a project, and to be able to reason about potential missing files (i.e., false negatives) more effectively. We describe multiple ontologies that enable the computation of SPGs while supporting an enterprise search engine. We demonstrate the novelty and application of this technique through an industrial case study, where a large engineering enterprise needs to make a long-term technological plan for large-scale document search, and has found the visualization-assisted approach to be more cost-effective than alternative approaches being studied.

show abstract

Semantic relation based personalized ranking approach for engineering document retrieval

Cited by 15 publications

References 30 publications

Deep Learning for Technical Document Classification

Deep Learning for Technical Document Classification

A Correlation-Experience-Demand Based Personalized Knowledge Recommendation Approach

Ontology-assisted provenance visualization for supporting enterprise search of engineering and business files

Contact Info

Product

Resources

About