Exploring users perceptions of NLP-based search systems for enhancing library accessibility in a public university of Bangladesh

Abstract

The study aims to examine users’ perceptions of the potential adoption of NLP-based search systems and explore how such systems could enhance library accessibility in a public university library in Bangladesh. A quantitative research methodology was employed, gathering data through surveys conducted among 387 respondents, including students and faculty members of NSTU. The survey utilized structured questionnaires with Likert-scale responses, dichotomous questions, and multiple-choice formats. The familiarity of digital Literacy and Natural language processing were analyzed through non-parametric tests like Mann-Whitney U and Kruskal-Wallis H. Statistical analysis tools, such as SPSS 27 and Excel 2021, were used for data processing and interpretation. The study revealed moderate familiarity with existing library information retrieval services, particularly among students. Respondents perceived that NLP-based search systems could improve search accuracy, relevance, and efficiency with 64.34% expressing positive expectations regarding their potential effectiveness. However, barriers like technical expertise, financial constraints, and user adaptation need to be addressed for successful implementation. The findings also highlight the need for continuous digital literacy training for both students and faculty. The research was limited to a single institution, potentially affecting its generalizability to other public university libraries in Bangladesh. This study suggests further exploration of NLP applications tailored to the complexities of the Bengali language for wider applicability and effectiveness.

Keywords

information retrieval services natural language processing artificial intelligence library services Bangladesh

Introduction

Background

Artificial intelligence (AI) is a rapidly emerging field that impacts various sectors, including education. Artificial Intelligence (AI) classified into various applicable forms, one of the well-known and most significant forms is “Natural language processing (NLP).” Natural language processing refers to the computational and mathematical modeling of human language, along with the design and development of systems capable of processing and analyzing linguistic data. Natural language processing has been applied in various fields, including information science, clinical research, computer science, text summarization, data and information retrieval. In the context of information storage and retrieval, NLP is used to preprocess documents, discover inter-term dependencies, and process user queries for effective searching.¹ Although NLP facilitates the efficient processing of large volumes of natural language search queries, its effectiveness may be constrained by ambiguity in user intent, domain-specific language variations, and contextual understanding, which can restrict the precision and depth of retrieved information.² Another challenge is the need for alternative approaches to collect information for analysis, which requires techniques to extract data from unstructured sources.³ Natural language processing forms the backbone of information retrieval, thereby enabling meaningful insights and actions to be extracted from complex structures in language and hence enabling artificial intelligence to make sense of digital information.⁴ These studies argue that information retrieval in Bangladesh’s public university libraries may be greatly enhanced by natural language processing (NLP). The Application of NLP in the field of information and communication technology (ICT) remains unexplored, despite its potential to optimize ICT processes. The execution of information retrieval technologies in academic libraries in Bangladesh is hindered by various factors, including inadequate infrastructure, a shortage of experienced personnel, financial constraints, and dependence on incomplete external services. Further hindering advancements in information availability are inadequate library collaboration and the high expense of digital preservation solutions.⁵

Stages involved with natural language processing

To properly understand, natural language processing goes through multiple stages. These stages include Figure 1.

With words serving as the fundamental unit of analysis, lexical analysis in natural language processing (NLP) employs techniques such as tokenization, stemming, and normalization to produce normalized word forms. If all lemmatized words are present, lexicons can be to identify base forms.⁶ To fully understand phrases and clauses, syntactic analysis examines word relationships and sentence structures, identifying parts of speech and sentence construction principles.⁷ By capturing objects and actions often using frames to depict events or scenarios semantic analysis can extract the deeper meaning of the text. Discourse analysis examines linguistic features such as anaphora, which influence word usage and idea references, thereby altering information retrieval. By interpreting the intended meaning of a document, pragmatic analysis can assist with tasks such as sentiment analysis, which classifies material based on positivity or negativity and infers mood (Figure 1).

Study purpose

The primary purpose of this study is to examine the potential use of natural language processing (NLP) and evaluate the current state of information retrieval services in public university libraries in Bangladesh. To achieve the milestone, the researcher determined following research objectives:

I. To identify the current state of information retrieval services in public university library.

II. To explore the potentials of NLP in information retrieval services in public university library.

III. To assess users’ perceptions regarding the potential effectiveness of NLP-based information retrieval services in improving user access to information.

Research questions

Considering the objectives, there are few questions that address the research problem:

RQ1: What are the current information retrieval services offered by a public university library?

RQ2: What are the challenges faced by library professionals in providing effective information retrieval services?

RQ3: What are users’ perceptions regarding the potential effectiveness of NLP-based information retrieval services in a public university library?

The significance of this study lies in the fact that it explores the possibility of using NLP technology for transforming the entire concept of information retrieval at public university libraries in Bangladesh. Furthermore, this study will also highlight existing deficiencies in library services and provide ways to address them, which could be related to poor digital literacy and lack of technical expertise. This study will be helpful for both the users and policymakers and library managers to implement technological measures for managing academic resources effectively. Public universities, like the Noakhali Science and Technology University in Bangladesh, confront considerable challenges when trying to deliver effective information retrieval services. These challenges range from infrastructural issues to the lack of efficient tools to facilitate natural language searches. Consequently, users find it difficult to retrieve pertinent information due to these limitations, which adversely affects their academic performance. The use of NLP technology has great potential in optimizing information retrieval operations; however, a few studies have been conducted to explore this application in the mentioned context. The study is confined to a single university, limiting its generalizability to other public universities in Bangladesh. Implementing NLP requires substantial financial investment and technical expertise, which may not be readily available in all university libraries.

Literature review

Emergence of information retrieval service

Information Retrieval (IR) refers to the process of retrieving and presenting information from large databases according to user queries. There are various steps involved in the process of Information Retrieval, including preprocessing, indexing, and ranking, which help to ensure timely and accurate information provision.⁸ Automated information retrieval systems need to be developed to enable the effective handling of increased volumes of information within business and academic contexts.⁹ Contemporary information retrieval systems involve the management of information organization, storage, search, and retrieval.¹⁰ There have been notable advances in digital retrieval systems used in libraries, increasing their effectiveness and accessibility. Digital retrieval systems make the operations within libraries more efficient and accessible through digital conversion of various media forms. Digital retrieval systems facilitate searches, including full-text searches, image searches, and multimedia searches. Machine learning enables effective data mining. The system provides personalization services according to user preferences and needs security measures from any unauthorized access.¹¹

Information retrieval (IR) with artificial intelligence (AI)

The development of artificial intelligence (AI) has brought about several innovations in the field of information retrieval (IR), especially when it comes to natural language processing, which enhances semantic comprehension of both the query and the documents. One such system, developed by Liu and Cheng in 2009,¹² is an intelligent IR system that enhances the effectiveness of information retrieval by adding context to the documents. AI-driven systems, especially ones involving NLP, have been shown to have a better ability at carrying out complex IR operations and analyzing texts comprehensively.¹³ Artificial Intelligence role in information retrieval (IR) Systems: Enhancing retrieval speed, accuracy, and management of multimedia content using AI in IR Systems.^14,15 Challenges of Using AI in Information Retrieval.^14–16

Natural language processing and artificial intelligence (AI) in modern information retrieval

Natural language processing is the field dedicated to empowering computers to understand, interpret, and produce human language in a meaningful and advantageous way. NLP involves the creation of algorithms and models that allow computers to analyze and comprehend natural language input.¹⁷ The primary objective of NLP is to create computational models capable of comprehending, interpreting, and generating human language. Zhou and Zhang⁷ propose a framework that applies in natural language processing techniques to information retrieval (IR) tasks. The paper discusses the challenges and limitations of implementing NLP in IR. The paper identifies the gaps and directions for future research. Russell-Rose and Stevenson¹⁵ provide an overview of the applications in information retrieval, that is, query expansion, document clustering and text summarization. Alhawiti¹⁸ analyzes the role of natural language processing in automated data retrieval, automated question answering, and text structuring and emphasizes the importance of NLP in indexing document collections and generating descriptions that represent the content of each document. Kumar et al.¹⁹ proposed a text-based image retrieval system that uses preprocessing, feature extraction, document clustering to retrieve relevant images based on text queries. Kowsher et al.²⁰ identify NLP has diverse applications across industries. Key applications include sentiment analysis for understanding emotions in text, text classification for organizing data, machine translation for translating languages, Natural language processing in the education and healthcare fields to classify unstructured data and enhance various aspects such as patient identification, healthcare perception analysis, student academic success, reading comprehension, and fairness of student evaluations.²¹ Shevendrakumar²² focuses on natural language processing (NLP) techniques to organize and find news articles from a large Reuters dataset, with the aim of helping journalists and researchers identify patterns and extract useful information. This study acknowledges some drawback like unclear categorization and a complex dataset, which may have affected the accuracy of clustering and retrieval results. Durga et al.²³ highlight the potential of using suitable text mining techniques to extract important information from text message, time and effort required for retrieving relevant information. The unstructured text data present challenges in structuring and classifying the content of the text data that may require sophisticated methods for analysis. Additionally, unstructured text data improve retrieval efficiency through semantic searches by analyzing the meaning of the query terms used. Unstructured text data provide several tools for enhancing content retrieval. Nabankema²⁴ seeks to explore and assess the influence of NLP techniques in enhancing retrieval efficiency. Malak and Ogurek²⁵ faced challenges in Information Retrieval (IR) tasks, especially in texts written in professional language. The challenge involves variations in the language vocabulary and grammar used. Moreover, the need for context in accurately retrieving content is another challenge. Ardehkhani et al.²⁶ identify some of the challenges faced in NLP for information retrieval, including language ambiguity, complex text structuring, and lack of adequate data on some languages.

Materials and methods

Research design

This study applied a quantitative approach to examine the potential use of natural language processing (NLP) as well as identify the current state of information retrieval services in a public university library in Bangladesh. “Quantitative research is described as the systematic analysis of phenomena through the collection of measurable data, utilizing statistical, mathematical, or computational methods. Quantitative research gathers data from current and prospective populations through sampling techniques and the distribution of online surveys, polls, and questionnaires, with results expressed numerically.”²⁷ At the time of this study, Noakhali Science and Technology University (NSTU) Library did not have an operational NLP-based information retrieval system. Consequently, respondents were not asked to evaluate an existing NLP application. Instead, the survey assessed users’ perceptions, expectations, and readiness regarding the potential implementation of NLP-based search systems based on their familiarity with existing library search services and the information provided in the questionnaire.²⁸

Research site

There are Fifty-five public Universities in Bangladesh (University Grant Commission, 2025). Amidst Noakhali Science and Technology University (NSTU) is one of them. The researcher selects NSTU as the research site because of its strategic position among the newly emerging public universities of Bangladesh with the highest importance attached to science and technology education.

Sample and sampling technique

The researcher conducted a survey using both printed and online methods. Out of the 350 students invited to participate, 265 completed the survey, resulting in a response rate of 75.71%. For the faculty members, 160 were invited, and 122 responded, yielding a response rate of 76.25%%. To conduct this study, a non-probability convenience sampling technique was employed to collect data from students and faculty members of Noakhali Science and Technology University. A non-probability convenience sampling technique was employed because of the ease of access to respondents and time limitations associated with data collection. Although convenience sampling facilitated efficient data gathering from active library users, the technique may introduce sampling bias and limit the generalizability of the findings beyond the selected institution. Therefore, the results should be interpreted cautiously, particularly when extending conclusions to public university libraries context in Bangladesh.

Data processing, analysis, and presentation

The survey responses have been collected and analyzed using statistical analysis software such as IBM SPSS 27 version and visualize the figures by Microsoft Excel 2021. Since the collected data were primarily ordinal in nature and did not satisfy the assumptions of normal distribution required for parametric analysis, non-parametric statistical tests were used. These tests like Mann-Whitney U and Kruskal-Wallis H. The use of non-parametric methods enabled the study to evaluate differences and relationships among variables without relying on strict distributional assumptions. The interpretation of results was based on significance values, rank distributions, and observed response patterns to assess users’ perceptions regarding NLP-based information retrieval services.

Hypothesis of the study

1. Null Hypothesis H₀1: There is no significant relationship between students’ education level and the self-rated literacy of digital technologies applied in libraries.

2. Null Hypothesis H₀2: Education level does not have a significant impact on students’ familiarity with the term “Natural Language Processing (NLP).

3. Null Hypothesis H₀3: Faculty member education level does not make any difference on their literacy of digital technologies applied in libraries.

4. Null Hypothesis H₀4: There is no significant association between faculty members education level and familiarity with the term “Natural Language Processing (NLP)”

Results

Demographic information of the respondents

The demographic Table 1 deliver insights about the respondents, comprising 265 students and 122 faculty members. Among students, 55.8% are male and 44.2% female, with most aged between 21 and 26 years (84.9%). The majority (44.5%) are Master’s students, while others span undergraduate levels. Additionally, 59.6% of students reside on campus, reflecting a preference for on-campus living facilities among students in higher education. For faculty members, 64.8% are male and 35.2% female, showing a noticeable gender gap. Most faculty members hold a Master’s degree (68%), followed by PhD holders (21.3%), with designations ranging across Lecturer (20.5%), Assistant Professor (41.8%), Associate Professor (24.6%), and Professor (13.1%). This highlights a well-qualified academic staff, with a significant proportion holding advanced qualifications. The data indicates a balanced representation of students and faculty in terms of gender, education levels, and professional roles, but with a need to address the gender disparity in faculty.

Table 1.

Demographic profile of the respondents.

Variables	Category	F	%
Students
Gender	Male	148	55.8
	Female	117	44.2
	Prefer not to say	0	0
Total		265	100
Age	18–21	20	7.5
	21–24	111	41.9
	24–26	114	43.0
	26 and above	20	7.5
Total		265	100
Level of education	1st Year	9	3.4
	2nd Year	41	15.5
	3rd Year	44	16.6
	4th Year	53	20.0
	Master’s	118	44.5
Total		265	100
Residential status	Resident	158	59.6
Residential status	Non-resident	107	40.4
Total		265	100
Faculty members
Gender	Male	79	64.8
	Female	43	35.2
	Prefer not to say	0	0
Total		122	100
Level of education	Bachelor’s	0	0
	Master’s	83	68.0
	M. Phil.	13	10.7
	PhD	26	21.3
Total		122	100
Designation	Lecturer	25	20.5
	Assistant Professor	51	41.8
	Associate Professor	30	24.6
	Professor	16	13.1
Total		122	100

N = Total Number of Respondents = 387.

Familiarity with the information retrieval services or searching system in library

Table 2 shows that familiarity with Information Retrieval Services in the library varies among respondents. A significant portion, 42.9%, is moderately familiar, while 26.4% are very familiar, and 6.5% are extremely familiar. However, 18.6% are slightly familiar, and 5.7% are not familiar at all. This indicates that most people have at least some levels of familiarity with these services.

Table 2.

Familiarity with the library information retrieval services.

Category	Frequency (F)	Percentage (%)
Not at all familiar	22	5.7
Slightly familiar	72	18.6
Moderately familiar	166	42.9
Very familiar	102	26.4
Extremely familiar	25	6.5

Participation in training programs or course work on information retrieval services

Figure 2 indicates that a significant majority of respondents (73.9%) have participated in training programs or coursework related to information retrieval services. Adversely, a smaller proportion (26.1%) have not had such exposure. This suggests that most individuals in the sample have engaged in some form of structured training in information retrieval, which may reflect a growing emphasis on developing knowledge in this area to meet the demands of modern library services and technological advancements.

Information searching experience in library

Table 3 summarizes individuals’ overall experiences with searching for information in the library. The majority, 29.2%, rated their experience as “Good,” followed by 28.4% who rated it as “Very good.” 17.3% was chosen as “Excellent,” while 16.0% rated their experience as “Fair.” Only 9.0% described their experience as “Poor.” This indicates that while many users find their library search experience satisfactory, there is room for improvement to enhance user satisfaction.

Table 3.

Overall experience with library search systems.

Category	Frequency (F)	Percentage (%)
Excellent	67	17.3
Very good	110	28.4
Good	113	29.2
Fair	62	16.0
Poor	35	9.0

Library staff assistance regarding information retrieval services

Figure 3 reveals the extent to which individuals have received assistance from library staff regarding information retrieval. The majority of library users, 76.5% have received assistance from library staff in retrieving information. In contrast, 23.5% indicated that they had not received such support, this highlights the significant role library staff play in aiding users with information retrieval.

Importance of the following aspects of information retrieval services

Table 4 portrays the importance of various aspects of information retrieval services, using a Likert scale. Among the factors, “speed of retrieving information” emerges as the most important (mean: 4.03, SD: 0.923), showing a strong consensus on its significance. Similarly, “availability of up-to-date resources” (mean: 3.99, SD: 1.154) ranks highly, though with slightly more variation in opinions. “User-friendly search interfaces” (mean: 3.94, SD: 0.954) and “access to a wide range of databases” (mean: 3.87, SD: 0.909) are also rated as important, with moderate consistency in responses. “Availability of library staff assistance,” while still considered important (mean: 3.73), has the highest variability (SD: 0.989), suggesting mixed opinions among users. Overall, users place a high priority on efficiency, accessibility, and modern resources, with slightly less emphasis on staff involvement.

Table 4.

Importance of information retrieval service attributes.

Statements	1 (%)	2 (%)	3 (%)	4 (%)	5 (%)	Mean (SD)
Availability of up-to-date resources	14 (3.6)	39 (10.1)	57 (14.7)	102 (26.4)	175 (45.2)	3.99 (1.154)
Access to a wide range of databases	5 (1.3)	30 (7.8)	69 (17.8)	191 (49.4)	92 (23.8)	3.87 (0.909)
User-friendly search Interfaces	5 (1.3)	23 (5.9)	89 (23.0)	143 (37.0)	127 (32.8)	3.94 (0.954)
Availability of library staff assistance	4 (1.0)	51 (13.2)	77 (19.9)	167 (43.2)	88 (22.7)	3.73 (0.989)
Speed of retrieving information	5 (1.3)	17 (4.4)	78 (20.2)	149 (38.5)	138 (35.7)	4.03 (0.923)

Note. 1 = Not at all Important, 2 = Slightly important, 3 = Moderately important 4 = Very important, 5 = Extremely important.

Evaluating literacy of digital technologies applied in libraries

Table 5 analyzes the digital literacy among different demographics. For students, no significant gender differences are observed (p = 0.945), but digital literacy significantly decreases as they progress academically, with 1st-year students having the highest mean rank (183.56) and Master’s students the lowest (110.47) (p = 0.000). Among faculty, gender differences are negligible (p = 0.965), and education level (Master’s, MPhil, PhD) does not significantly impact digital literacy (p = 0.761). Remark: While faculty demonstrates consistent digital literacy, students show a noticeable decline in proficiency as they advance academically, suggesting a need for sustained focus on digital skill development throughout their education.

Table 5.

Digital technological literacy by different demographic library users.

Variables	Gender/Education level	N	Mean rank	Mann-Whitney U	Kruskal-Wallis H	Wilcoxon W	Z	Sig. Value	P-value
Students’ Digital Literacy	Male	148	133.28				−
Students’ Digital Literacy	Female	117	132.65	8617.0		15,520.0	0.070		0.945
Faculty’s Digital Literacy	Male	79	61.59				−
Faculty’s Digital Literacy	Female	43	61.33	1691.0		2637.0	0.044		0.965
Students’ Digital Literacy	1st Year	9	183.56
	2nd Year	41	175.46
	3rd Year	44	134.32		30.092			0.000
	4th Year	53	140.63
	Master’s	118	110.47
Faculty’s Digital Literacy	Master’s	83	60.05
	M.Phil.	13	63.38		0.546			0.761
	PhD	26	65.19

Familiarity with natural language processing

Table 6 explores the familiarity with natural language processing (NLP) across different demographics. For students, there is no significant difference in NLP familiarity between male and female groups (Mann-Whitney U = 8337.0, p = 0.587). Similarly, for faculty members, no significant difference is observed between male and female participants (Mann-Whitney U = 1649.0, p = 0.783). However, among students across academic years, there is a significant difference in familiarity (Kruskal-Wallis H = 36.851, p = 0.000), indicating variability by year of study. For faculty members, no significant difference in familiarity is found between M.Phil. and PhD holders (Kruskal-Wallis H = 1.183, p = 0.553). The analysis indicates that gender does not significantly influence familiarity with NLP among students or faculty members. However, there is a notable difference in familiarity levels among students across academic years, suggesting that familiarity increases with progression in education. On the other hand, faculty members’ educational qualifications (M.Phil vs PhD) do not show a significant impact on their NLP familiarity.

Table 6.

Familiarity with natural language processing (NLP) by different demographic library users.

Variables	Gender/education level	N	Mean rank	Mann-Whitney U	Kruskal-Wallis H	Wilcoxon W	Z	Sig. value	P-value
NLP Familiarity-Student	Male	148	130.83	8337.0		19,363.0	−		0.587
NLP Familiarity-Student	Female	117	135.74				0.544
NLP Familiarity-Faculty	Male	79	60.87	1649.0		4809.0	−
NLP Familiarity-Faculty	Female	43	62.65				0.275		0.783
NLP Familiarity-Students	1st Year	9	65.00		36.851			0.000
	2nd Year	41	99.15
	3rd Year	44	110.98
	4th Year	53	128.98
	Master’s	118	159.97
NLP Familiarity-Faculty	Master’s	83	60.67		1.183			0.553
	M.Phil.	13	55.42
	PhD	26	67.17

Potential advantages in integrating NLP technologies into library’s information retrieval system

Table 7 reveals that respondents find several benefits in using NLP-based search tools. The most recognized benefit was enhanced search accuracy (29.9%), followed by personalized user experience (19%) and effective handling of unstructured data (17.2%). Language processing capabilities (17.6%) and improved relevance feedback (16.3%) were also acknowledged as significant advantages. These findings underscore the perceived advantages of NLP tools in improving search precision, user experience, and data management.

Table 7.

Perceived advantages of NLP-based search tools.

Factors	Frequency (F)	Percentage (%)
Enhanced search accuracy	299	29.90
Effective handling of unstructured data	172	17.20
Personalized user experience	190	19.00
Language processing capabilities	176	17.60
Improved relevance feedback	163	16.30
Other	0	0

Potential challenges in integrating NLP technologies into library’s information retrieval system

Table 8 highlights several challenges associated with implementing NLP-based search tools. The main challenges identified in integrating NLP technologies into library systems were technical expertise (25.28%) and financial constraints (20.06%). Data security (17.54%) and user adaptation/training (15.39%) were also significant, while interoperability (12.31%) and ethical considerations (9.42%) were less prevalent concerns. These findings suggest that while there are multiple barriers to adoption, technical expertise and financial constraints are the most pressing.

Table 8.

Potential challenges in implementing NLP technologies.

Factors	Frequency (F)	Percentage (%)
Financial constrains	215	20.06
Technical expertise	271	25.28
Data security	188	17.54
Interoperability	132	12.31
User adaptation and training	165	15.39
Ethical considerations	101	9.42
Other	0	0

Frequency of using the university library’s search system

Table 9 indicates varying frequencies of using the university library’s search system among respondents. Among 387 respondents, 37.2% use the university library’s search system monthly, followed by 30.7% who use it rarely, and 22.0% who use it weekly. Only 4.4% of respondents use it daily, while 5.7% never use the system. This suggests that while the system is utilized by some on a regular basis, a large number of respondents use it infrequently or not at all.

Table 9.

Usage frequency of the university library search system.

Category	Frequency (F)	Percentage (%)
Daily	17	4.4
Weekly	85	22.0
Monthly	144	37.2
Rarely	119	30.7
Never	22	5.7

Efficiency of NLP-based retrieval methods compared to traditional techniques

Figure 4 indicates the efficiency of NLP-based retrieval methods compared to traditional techniques, 64.34% of respondents agreed that NLP-based methods would be more effective than traditional search systems upon implemented. While 19.38% disagree, and 16.28% are unsure. This suggests a strong preference for NLP-based methods among users, though a significant portion remains uncertain about their efficiency.

User perspectives on NLP-enhanced library services

Table 10 indicates that users generally perceive the NLP-based information retrieval service positively. Key benefits include improved accuracy and relevance of search results (mean: 3.98, SD: 1.108), ease of finding information quickly and efficiently (mean: 3.89, SD: 0.827), and an enhanced overall search experience (mean: 3.77, SD: 0.865). Respondents also acknowledged the services’ ability to uncover new and relevant information (mean: 3.76, SD: 0.967), reduce research effort (mean: 3.65, SD: 1.106), improve access to authoritative and up-to-date sources (mean: 3.76, SD: 1.007), and boost confidence in the quality and reliability of information (mean: 3.74, SD: 0.987). These findings highlight the value users associate with NLP-based enhancements.

Table 10.

Users’ perceptions of potential NLP-enhanced library services.

Statements	1 (%)	2 (%)	3 (%)	4 (%)	5 (%)	Mean (SD)
NLP-based information retrieval service has improved the accuracy and relevance of search results	14 (3.6)	26 (6.7)	78 (20.2)	103 (26.6)	166 (42.9)	3.98 (1.108)
NLP-based service has made it easier for me to find the information I need quickly and efficiently	5 (1.3)	20 (5.2)	66 (17.1)	218 (56.3)	78 (20.2)	3.89 (0.827)
NLP-based service has enhanced my overall searching experience	4 (1.0)	12 (3.1)	139 (35.9)	146 (37.7)	86 (22.2)	3.77 (0.865)
NLP-based service has helped me discover new and relevant information that I would not have found otherwise	4 (1.0)	47 (12.1)	73 (18.9)	177 (45.7)	86 (22.2)	3.76 (0.967)
NLP-based service has reduced the time and effort required to conduct research and complete assignments	23 (5.9)	32 (8.3)	95 (24.5)	146 (37.7)	91 (23.5)	3.65 (1.106)
NLP-based service has improved my ability to find and access up-to-date and authoritative sources of information	9 (2.3)	39 (10.1)	81 (20.9)	163 (42.1)	95 (24.5)	3.76 (1.007)
NLP-based service has increased my confidence in the quality and reliability of the information	13 (3.4)	26 (6.7)	95 (24.5)	167 (43.2)	86 (22.2)	3.74 (0.987)

Note. 1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree.

Discussion

The findings from the study reflect the RQ1 that while the information retrieval services (IRS) in public university libraries are functional, there is a clear need for improvement in terms of usage and familiarity. As shown in Table 2, a significant portion of respondents, particularly students, reported moderate to low familiarity with these services. This lack of familiarity can likely be attributed to the absence of formal training programs, as shown in Figure 2, where 26.1% of respondents indicated they had never participated in any related training. In contrast, faculty members, as reflected in Table 2, demonstrated greater familiarity and more consistent engagement with IRS, likely due to their academic and research needs. Underscore the importance of targeted awareness campaigns and user-specific training programs. This disparity suggests that while the system functions, it requires more inclusive outreach and educational initiatives to ensure all users, particularly students, can maximize its potential. This discussion mostly aligns with a previous study authored by Du and Evans²⁹ stated while university library services such as Information retrieval and document delivery are widely recognized and utilized by academic users, particularly among students. It was noted that awareness of library services and confidence in using them grows as students’ progress through their education experience raise. However, the study also found that many students lacked participation in workshops or training sessions, which aligns with the challenges of not being familiar with digital technology applied in academic libraries. Conversely, faculty members are more frequently seek out library resources, reflecting a greater familiarity and confidence in utilizing these services.

The RQ2 was associated to the challenges faced by library professionals in terms providing effective information retrieval services and the findings suggest that NLP has significant potential to enhance the effectiveness and efficiency of information retrieval services in public university libraries. As indicated in Table 7, respondents recognized key advantages of NLP, such as enhanced search accuracy (29.9%) and the ability to process complex queries with greater efficiency (19.0%). These benefits suggest that NLP could play a critical role in meeting the evolving needs of library users, improving both the quality of search results and user satisfaction. However, Table 8 reveals challenges in implementing NLP systems. The most significant barriers identified were a lack of technical expertise among staff (25.28%) and financial constraints (20.06%). These findings emphasize that, although the potential of NLP is widely recognized, its successful implementation depends on overcoming these barriers through adequate investments in training programs and securing the necessary resources for system integration. Libraries need to prioritize both technical capacity building and financial support to fully leverage NLP’s capabilities and enhance their information retrieval systems. This discussion also supported by Taskin and Al³⁰ and Wolfram.³¹ They stated the integration of NLP techniques into information retrieval services in university libraries holds significant promise. Studies have demonstrated the potential for NLP to enhance search accuracy and relevance, which can greatly benefit library patrons.³⁰ However, a key challenge is the lack of awareness and familiarity with NLP among students, faculty, and even library professionals.³¹ Despite this, both faculty and library staff have expressed openness to adopting NLP-based systems, recognizing the advantages they can offer.³⁰

The RQ3 was related to measuring effectiveness of NLP-based IRS systems in public university libraries was evaluated positively by the majority of respondents. As seen in Figure 4, 64.34% of users found NLP-based methods more effective than traditional search techniques. The enhanced search accuracy (mean score: 3.98) and faster retrieval times (mean score: 3.89) reported by respondents in Table 10 demonstrate that respondents believe NLP-based search systems have the potential to improve user experience by reducing search effort and increasing the relevance of search results. However, these results suggest that while NLP-based systems could provide several advantages over traditional search systems. There is still room for improvement in terms of search accuracy and reliability. Addressing these issues will be key to maximizing the effectiveness of NLP systems and ensuring they meet the diverse needs of users consistently. Despite these challenges, the overall positive feedback indicates that NLP holds considerable promise for transforming library information retrieval systems, with continued refinement needed to fully realize its potential. The effectiveness of NLP-based information retrieval services in university libraries has been evaluated in various studies. While some studies have shown positive results, such as increased user satisfaction with NLP-based tools, others have reported mixed findings, highlighting the need for further research and development.³⁰

Research gaps and future directions

Although there have been many developments in AI-NLP-based information retrieval systems, current research has focused more on the technical side rather than from the perspective of end users. The problems of linguistics, lack of structured database, context and the challenges posed by languages like Bengali have not been investigated yet. Another area that requires more attention is to study the views, knowledge and preparedness of Bangladeshi academic librarianship towards digital and AI-based information retrieval system. Researcher will need to develop algorithms and techniques of NLP especially for Bengali language keeping its unique linguistic properties in mind. After reading about the scope and limitations of this study, users and librarians of academic libraries of Bangladesh can use it as the starting point to develop their thoughts on information search systems in the library. This study may also have an impact on decisions made by library managers.

Conclusions and recommendation

This study seeks to explore the possibility of employing NLP through user’s perception in enhancing the information retrieval facilities at the university library in Bangladesh. Although the current systems have demonstrated some level of functionality, the application of NLP would further enhance the accuracy, speed, and performance of information retrieval in such systems. However, issues like technical limitations, financial limitations, and user adoption require a deliberate effort by way of conducting training programs and allocating resources accordingly. This study provides insights into the need to modernize library facilities in accordance with the requirements of library users, hence offering policy makers and administrators useful advice. In the future, the application of NLP should be pursued despite the difficulties in dealing with languages like Bengali that have complex phonetics and varied forms of expression.

Footnotes

Acknowledgement

The author affirms that all the ideas and interpretations presented in this study are entirely author’s own. However, he also acknowledges that he used the free version of ChatGPT (GPT 4.0) to check clarity and flow, readability, and grammatical errors.

ORCID iD

Jahid Bin Sultan

Ethical considerations

This study was conducted in accordance with standard research ethics principles. Participation in this study was voluntary, and respondents were informed of the purpose of the research before data collection. Informed consent was obtained from all participants. No personal identifiable information was collected, and all responses were recorded anonymously to ensure confidentiality.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Author biography

Jahid Bin Sultan is a young information professional. He is currently working at the Military Institute of Science and Technology (MIST) under the Supervision of the Research and Development Wing. He completed both his Bachelor’s and Master’s degrees in Information Science and Library Management from Noakhali Science and Technology University (NSTU). Jahid is passionate about learning new technological advancements in the LIS field, with research interests in digital information systems, AI in libraries and technology-driven information services.

References

Khan

Bhojan

Nunavathu

. Cross-language information retrieval. International Journal of Research Publication and Reviews, 2022, pp. 1612–1618. https://doi.org/10.55248/gengpi.2022.3.6.13

Moon

Carlson

Moser

, et al. Identifying information gaps in electronic health records by using natural language processing: gynecologic surgery history identification. J Med Internet Res 2022; 24: e29015. https://doi.org/10.2196/29015

Bai

Guo

, et al. Enhanced natural language interface for web-based information retrieval. IEEE Access 2020; 9: 4233–4241. https://doi.org/10.1109/ACCESS.2020.3048164

Pattanayak

. Natural language processing. In: Pro Deep Learning with TensorFlow 2.0. Springer, 2023, pp. 293–405. https://doi.org/10.1007/978-1-4842-8931-0_4

Alam

. Digitization and developing digital library and information support in Bangladesh: professional challenges. Bangladesh Journal of Library and Information Science 2012; 2(1): 57–66. https://doi.org/10.3329/BJLIS.V2I1.12922

Smeaton

. Progress in the application of natural language processing to information retrieval tasks. Comput J 1992; 35(3): 268–278. https://doi.org/10.1093/comjnl/35.3.268

Alizadeh

Fattahi

Davarpanah

. Applying natural language processing techniques for effective Persian-English cross-language information retrieval. Int J Integrated Supply Manag 2012; 8(2): 89–98, Available from. https://profdoc.um.ac.ir/articles/a/1016924.pdf

Levine-Clark

Carter

. ALA glossary of library and information science. 4th ed. American Library Association, 2013.

Larson

. Information retrieval systems. In: Encyclopedia of Library and Information Sciences. 3rd ed, 2010, pp. 2553–2563.

10.

Sonawane

Mahalle

Ghotkar

. Information retrieval and natural language processing. In: Studies in big data. Springer, 2022. https://doi.org/10.1007/978-981-16-9995-5

11.

Wendi

. Research on the application of computer digital retrieval technology in the construction of library and information database. In: 2022 International Conference on Applied Physics and Computing (ICAPC). IEEE, 2022, pp. 219–223. https://doi.org/10.1109/ICAPC57304.2022.00047

12.

Liu

Cheng

. Artificial intelligent information retrieval using assigning context of documents. In: 2009 International Conference on Networks Security, Wireless Communications and Trusted Computing. IEEE, 2009, pp. 592–595. https://doi.org/10.1109/NSWCTC.2009.374

13.

Boughanem

Akermi

Pasi

, et al. Information retrieval and artificial intelligence. In: A Guided Tour of Artificial Intelligence Research. : Springer, 2020, vol III, pp. 147–180. https://doi.org/10.1007/978-3-030-06170-8_5

14.

Huang

. Research on the application of artificial intelligence in image and text database retrieval. Frontiers in Computing and Intelligent Systems 2022; 2(1): 39–41. https://doi.org/10.54097/fcis.v2i1.2708

15.

Russell-Rose

Stevenson

. The role of natural language processing in information retrieval. In: Information Retrieval: Searching in the 21st Century, 2009, pp. 215–231. https://doi.org/10.1002/9780470033647.ch10

16.

Ceri

Bozzon

Brambilla

, et al. An introduction to information retrieval. In: Web Information Retrieval. Springer, 2013, pp. 3–11. https://doi.org/10.1007/978-3-642-39314-3_1

17.

Chowdhury

. Natural language processing. Annu Rev Inf Sci Technol 2003; 37(1): 51–89. https://doi.org/10.1002/aris.1440370103

18.

Alhawiti

. Role of natural language processing in information retrieval: challenges and opportunities. International Journal of Computer and Information Engineering 2015; 8(12): 3658–3662, Available from. https://publications.waset.org/10000118/pdf

19.

Kumar

PMA

Rao

TSM

Raj

, et al. An efficient text-based image retrieval using natural language processing techniques. In: Advances in Computational Intelligence and Communication Technology. Springer, 2021. https://doi.org/10.1007/978-981-15-5400-1_52

20.

Kowsher

Uddin

Tahabilder

, et al. BanglaLM: data mining based Bangla corpus for language model research. In: 2021 Third International Conference on Inventive Research in Computing Applications (ICIRCA). IEEE, 2021. https://doi.org/10.1109/ICIRCA51532.2021.9544818

21.

Kanaparthi

. Examining natural language processing techniques in the education and healthcare fields. Int J Eng Adv Technol 2022; 12(2): 8–18. https://doi.org/10.35940/ijeat.B3861.1212222

22.

Shevendrakumar

. Clustering and retrieval of news articles using natural language processing. Indian Sci J Res Eng Manag 2023; 7(6). https://doi.org/10.55041/ijsrem24154

23.

Durga

Sanjana

Baig

, et al. Information extraction from text messages using natural language processing. In: 2023 International Conference on Computer Communication and Informatics (ICCCI). IEEE, 2023, pp. 1–6. https://doi.org/10.1109/ICCCI56745.2023.10128641

24.

Nabankema

. Evaluation of natural language processing techniques for information retrieval. European Journal of Information and Knowledge Management 2024; 3(1): 38–49. https://doi.org/10.47941/ejikm.1752

25.

Malak

Ogurek

. Including natural language processing and machine learning into information retrieval. Computer Science & Information Technology 2019: 13–18. https://doi.org/10.5121/csit.2019.91202

26.

Ardehkhani

Vahedi

Aghababa

. Challenges in natural language processing and natural language understanding by considering both technical and natural domains. In: 2023 6th International Conference on Pattern Recognition and Image Analysis (IPRIA). Tehran: IEEE, 2023, pp. 1–6. https://doi.org/10.1109/IPRIA59240.2023.10147185

27.

Habiba

. Towards the establishment of digital library: an investigation into the current status of university libraries in Bangladesh [PhD thesis]. University of Dhaka, 2020.

28.

Zhou

Zhang

. NLPIR: a theoretical framework for applying natural language processing to information retrieval. J Am Soc Inf Sci Technol 2003; 54(2): 115–123. https://doi.org/10.1002/asi.10193

29.

Evans

. Academic library services support for research information seeking. Aust Acad Res Libr 2011; 42(2): 103–120. https://doi.org/10.1080/00048623.2011.10722217

30.

Taskin

. Natural language processing applications in library and information science. Online Inf Rev 2019; 43(4): 676–690. https://doi.org/10.1108/OIR-07-2018-0217

31.

Wolfram

. Bibliometrics, information retrieval and natural language processing: natural synergies to support digital library research. In: Proceedings of BIRNDL, 2016, pp. 6–13.