Exploring Vector-Based Methods for Effective Document Retrieval
DOI:
https://doi.org/10.61173/077sp881Keywords:
Vector Searching, Document Indexing, Semantic Search, LLM, Word Embedding, Information RetrievalAbstract
With the development of online storage services and the internet, people, especially students, often collect numerous documents from the servers. Searching from abundant documents can become a difficult task. With the rapid development of the LLM (Large Language Model) and word embedding technique, people should have the chance to find the information in a new way, which is much more effective. In this paper, with the vectorization of documents and users’ questions with different models, they can somehow understand the words of users and find the correspond documents which the users want. And LLM can directly extract the important information in the documents. Therefore, users can easily find the information they want from a large number of documents.. This paper provides four methods with three main models to convert documents to vectors. The best one can retrieve more than 90% documents in the testing data set with the given keywords.
References
[1] Harris, Z. S. (1954). Distributional structure. Word, 10(23), 146-162.
[2] Landauer, T. K., Foltz, P. W., & Laham, D. (1998). An introduction to latent semantic analysis. Discourse Processes, 25(2-3), 259–284.
[3] Brown, P. F., Desouza, P. V., Mercer, R. L., Della Pietra, V. J., & Lai, J. C. (1992). Class-based n-gram models of natural language. Computational Linguistics, 18(4), 467–479.
[4] 来斯惟 . (2016). 基于神经网络的词和文档语义向量表示 方法研究 (Doctoral dissertation, 北京 : 中国科学院大学 ).
[5] Church, K. W. (2017). Word2Vec. Natural Language Engineering, 23(1), 155- 162.
[6] Le, Q., & Mikolov, T. (2014,). Distributed representations of sentences and documents. In Proceedings of the International Conference on Machine Learning (pp. 1188-1196). PMLR.
[7] Zhu, W., Wu, X., & Chen, J. (2019). Spatial keyword query of region-of-interest based on the distributed representation of point-of-interest. ISPRS International Journal of Geo- Information, 8(6), 287.
[8] Yucel, O., & Sen, S. (2018). Language independent recommender agent. The Knowledge Engineering Review, 33, e15.
[9] de Paula, L. B., Villaça, R. D. S., & Magalhães, M. F. (2011). Organization of multimedia data for conceptual search based on ontologies. Journal of the Brazilian Computer Society, 17, 241- 254.
[10] Jiang, F., Liu, Y. Q., Luan, H. B., Sun, J. S., Zhu, X., Zhang, M., & Ma, S. P. (2015). Microblog sentiment analysis with emoticon space model. Journal of Computer Science and Technology, 30(5), 1120-1129.
[11] Moallim, A. A., & Yun, L. J. (2021). Structural Application of Medical Image Report Based on Bi-CNNs-LSTM-CRF. In Advances in Computer, Communication and Computational Sciences: Proceedings of IC4S 2019 (pp. 365-377). Springer Singapore.
[12] Xiong, Y., Chen, S., Qin, H., Cao, H., Shen, Y., Wang, X., ... & Tang, B. (2020). Distributed representation and one-hot representation fusion with gated network for clinical semantic textual similarity. BMC medical informatics and decision making, 20, 1-7.
[13] de Paula, L. B., Villaça, R. D. S., & Magalhães, M. F. (2011). Organization of multimedia data for conceptual search based on ontologies. Journal of the Brazilian Computer Society, 17, 241- 254.
[14] Jatnika, D., Bijaksana, M. A., & Suryani, A. A. (2019). Word2vec model analysis for semantic similarities in english words. Procedia Computer Science, 157, 160-167.
[15] Alzubi, J. A., Jain, R., Singh, A., Parwekar, P., & Gupta, M. (2023). COBERT: COVID-19 question answering system using BERT. Arabian journal for science and engineering, 48(8), 11003-11013.
[16] Pant, G., & Srinivasan, P. (2010). Predicting web page status. Information Systems Research, 21(2), 345-364.
[17] OpenAI. (2024). ChatGPT (Version GPT-4) [Large language model].
[18] Kimi. (n.d.). Kimi intelligent assistant [Online].
[19] Ma, L., & Zhang, Y. (2015, October). Using Word2Vec to process big text data. In 2015 IEEE International Conference on Big Data (Big Data) (pp. 2895-2897). IEEE.
[20] Aizawa, A. (2003). An information-theoretic perspective of tf–idf measures. Information Processing & Management, 39(1), 45-65. Appendixces I Nguyen:2007 @InProceedings{Nguyen:2007, author = {Thuy Dung Nguyen and Min-Yen Kan}, title = {Key phrase Extraction in Scientific Publications}, booktitle = {Proceeding of International Conference on Asian Digital Libraries}, year = {2007}, pages = {317--326} Dean&Francis Xian Zhang } Appendix II Models of MS MARCO Approach NDCG@10 (TREC DL 19 Reranking) MRR@10 (MS Marco Dev) Queries (GPU / CPU) D o c s ( G P U / CPU) Models tuned for cosine-similarity msmarco-MiniLM-L-6-v3 67.46 32.27 18,000 / 750 2,800 / 180 msmarco-MiniLM-L-12-v3 65.14 32.75 11,000 / 400 1,500 / 90 msmarco-distilbert-base-v3 69.02 33.13 7,000 / 350 1,100 / 70 msmarco-distilbert-base-v4 70.24 33.79 7,000 / 350 1,100 / 70 msmarco-roberta-base-v3 69.08 33.01 4,000 / 170 540 / 30 Models tuned for dot-product msmarco-distilbert-base-dot-prod-v3 68.42 33.04 7,000 / 350 1100 / 70 msmarco-roberta-base-ance-firstp 67.84 33.01 4,000 / 170 540 / 30 msmarco-distilbert-base-tas-b 71.04 34.43 7,000 / 350 1100 / 70 Appendix III Combination with LLM GPT-4o:[17] User: What is IEEE 802.11 in Bluetooth? GPT-4o: File: ../data/all45.txt Sentence: “This same frequency band is already in use by microwave ovens and the popular Wireless Local Area Network (WLAN) devices implementing the IEEE 802.11 standard specifications [8] .” Answer: IEEE 802.11 refers to the set of standards for Wireless Local Area Network (WLAN) devices, which share the same 2.4 GHz ISM band as Bluetooth devices. moonshot-v1-8k: [18] User: Wireless technology Kimi: File: ../data/all176.txt Sentence: “Recent advances in wireless networking technology, low-power digital circuits, sensing materials and Micro Electro-Mechanical Systems (MEMS) opened up the possibility of building small sensor devices capable of data processing, remote sensing and wireless communication.” Answer: The development in wireless technology, along with advancements in low-power digital circuits, sensing materials, and MEMS, has enabled the creation of compact sensor devices. These devices are capable of performing data processing, remote sensing, and wireless communication, which are crucial for the operation of wireless sensor networks.
Downloads
Published
Issue
Section
License
Copyright (c) 2025 by the authors.
This work is licensed under a Creative Commons Attribution 4.0 International License.
