Enhancing the Precision of Machine Learning in the Library Profession

1. Introduction to Machine Learning in Libraries

In recent years, the field of library and information science has been witnessing a significant shift towards the incorporation of advanced technologies to enhance the efficiency and effectiveness of library services. One such technology that has gained prominent recognition in the context of libraries is machine learning. Machine learning, a subset of artificial intelligence, offers libraries novel solutions to optimize resource management, streamline operations, and provide personalized services to users.

The application of machine learning in libraries is multifaceted, encompassing diverse areas that can revolutionize the way libraries operate and serve their users. One of the primary applications of machine learning in libraries is in data analysis and information retrieval. Libraries house extensive collections of physical and digital resources, along with valuable user data. Machine learning algorithms can be employed to analyse these datasets, identify patterns, and extract valuable insights that can inform decision-making processes related to collection development, resource allocation, and user engagement strategies (Tufail, Riggs, Tariq, & Sarwat, 2023). Harnessing machine learning capabilities, libraries can unlock the potential of their data resources to make informed and data-driven decisions that align with user needs and preferences.

Moreover, machine learning holds immense promise in enhancing user experiences and services within library settings. Personalized recommender systems, powered by machine learning algorithms, can analyse user behaviour, preferences, and interactions with library resources to deliver tailored recommendations and content suggestions (Das & Islam, 2021). These recommendation systems help users discover new materials, navigate library collections with ease, and engage with content relevant to their interests, thereby fostering a more enriching and personalized user experience. Additionally, machine learning can enable libraries to automate routine tasks such as cataloguing, classification, and inventory management, allowing library staff to focus on high-value activities that enhance user services and engagement.

Furthermore, the integration of machine learning in libraries has the potential to foster community engagement and lifelong learning initiatives. By leveraging natural language processing (NLP) and sentiment analysis, libraries can gain insights into user feedback, sentiment, and engagement levels with library services, enabling them to tailor communication strategies and outreach efforts to better connect with their communities (Bi, Wang, Zhang, Huang, Wu, Gong, & Ni, 2022). Additionally, machine learning can facilitate predictive analytics, enabling libraries to anticipate user needs, forecast trends, and adapt their services proactively to meet evolving demands and preferences of their users.

While the adoption of machine learning in libraries offers numerous benefits, it also poses ethical and privacy considerations that necessitate careful attention and deliberation. As libraries collect and analyse user data to improve services and personalize experiences, it is imperative to uphold principles of data privacy, transparency, and accountability (Shah A., Iqbal, Aziz, Rana, Khalid, Cheah, & Arif, 2022). Safeguarding user privacy and ensuring the ethical use of machine learning technologies require libraries to establish robust data governance frameworks, adhere to relevant data protection regulations, and engage in transparent communication with users regarding data practices.

Machine learning represents a transformative frontier for libraries, empowering them to leverage data-driven insights, automate processes, and provide user-centric services in alignment with the evolving digital landscape. Embracing machine learning technologies and adhering to ethical guidelines, libraries can position themselves as dynamic hubs of knowledge, innovation, and community engagement. As libraries navigate the digital age and seek to enhance their offerings, machine learning emerges as a pivotal tool that can enable libraries to adapt, innovate, and thrive amidst changing user expectations and technological advancements.

2. Applications of Machine Learning in Libraries

Machine learning technologies have emerged as powerful tools with the potential to transform traditional library services and operations. The integration of machine learning in libraries offers a wide range of applications that enhances resource management, user engagement, and decision-making processes. One of the key applications of machine learning in libraries is in data analysis and information retrieval. A study by Tufail, Riggs, Tariq, and Sarwat (2023) machine learning analyses vast amounts of library data, including collections, user interactions, and circulation patterns, to extract valuable insights that inform collection development, resource allocation, and user engagement strategies.

Furthermore, machine learning plays a crucial role in personalized recommender systems within library settings. These systems utilize algorithms to analyse user behaviour, preferences, and interactions with library resources, enabling libraries to offer tailored recommendations and content suggestions to users (Das & Islam, 2021). Through providing personalized recommendations, libraries enhances user experiences, increase user satisfaction, and promote the discovery of relevant materials from their collections.

Another important application of machine learning in libraries is the automation of routine tasks such as cataloguing, classification, and inventory management. Bi, Wang, Zhang, Huang, Wu, Gong, and Ni (2022) ague that machine learning automate time-consuming processes, allowing library staff to focus on more value-added activities that improve user services and engagement. Through automation, libraries streamline operations, reduce manual workload, and allocate resources more effectively, leading to enhanced efficiency and productivity.

Moreover, machine learning facilitate predictive analytics in libraries, enabling them to forecast user needs, anticipate trends, and adapt services proactively. Shah A., Iqbal, Aziz, Rana, Khalid, Cheah, and Arif (2022) identifies key predictive models of machine learning in libraries, libraries optimizes resource planning, tailor services to meet user demands, and enhance decision-making processes. Predictive analytics empower libraries to stay ahead of evolving user preferences and ensure that their services remain relevant and responsive to changing needs.

The applications of machine learning in libraries are diverse and impactful, spanning data analysis, personalized recommendations, automation, and predictive analytics. By harnessing the power of machine learning technologies, libraries can optimize their operations, engage users more effectively, and improve the overall library experience. As libraries continue to embrace digital transformation, machine learning stands out as a valuable tool that can drive innovation, efficiency, and user-centric services in the ever-evolving library landscape.

3. Challenges in Implementing Machine Learning in Library Operations

Implementing machine learning in library operations presents a transformative opportunity for libraries to enhance services and optimize workflows. However, this integration is not without its challenges, which can hinder the successful adoption and implementation of machine learning technologies in library settings. Understanding and addressing these challenges is essential to unlock the full potential of machine learning in libraries.

One of the key challenges in implementing machine learning in library operations is data quality and availability. According to Croce, Caroti, De Luca, Jacquot, Piemonte, and Véron, (2021) Libraries often encounter issues related to the quality, consistency, and completeness of their data, which can impact the effectiveness of machine learning algorithms. Inaccurate or incomplete data can lead to biased outcomes and erroneous predictions, undermining the reliability and utility of machine learning applications in libraries. Therefore, libraries must invest resources in data cleaning, validation, and enrichment to ensure that the data used for machine learning is of high quality and relevance.

Another significant challenge is the technical expertise required to develop, deploy, and maintain machine learning models in library operations. In Nigeria, Ajani, Tella, Salawu, and Abdullahi (2022) report that most librarians expressed fear that the use of AI in libraries could take over their jobs. However, in a study on A data-driven Machine Learning approach to creativity and innovation techniques selection in solution development, de Carvalho Botega and da Silva (2022), reported that Library staff lack the necessary skills and knowledge in data science, programming, and algorithm development to effectively leverage machine learning technologies. Training staff or hiring external expertise can be costly and time-consuming, posing a barrier to the successful implementation of machine learning solutions in libraries. Overcoming the skills gap and building capacity within library teams is crucial to building sustainable machine learning initiatives.

Furthermore, ethical and privacy considerations present complex challenges in the implementation of machine learning in library operations. Padilla (2019) was of the opinion that libraries handle sensitive user data and must ensure compliance with privacy regulations and ethical standards when implementing machine learning algorithms. Issues such as data protection, algorithm transparency, and user consent require careful attention to safeguard user rights and trust in library services powered by machine learning technologies.

Moreover, resistance to change and organizational culture can impede the adoption of machine learning in library operations. Georgalis, Samaratunge, Kimberley, and Lu (2015) reported that libraries are faced with internal resistance from staff members who are skeptical of new technologies or apprehensive about the implications of automation on their roles. Overcoming resistance to change requires effective communication, training, and leadership support to foster a culture of innovation and collaboration that embraces the potential of machine learning in enhancing library services. In their study on the potential role of a librarian in a machine dominated library environment, Wheatley and Hervieux (2019) reported that in the USA and Canada, a few institutions were participating in the creation of machine hubs and that there was generally a lack of awareness to the machine learning trends in academic libraries and that while research on machine learning had been exponentially increasing in other fields, this had not been the case in the discipline of information science.

Addressing the challenges of data quality, technical expertise, ethical considerations, and organizational culture is essential for successful implementation of machine learning in library operations. By recognizing and mitigating these challenges, libraries can harness the power of machine learning to improve resource management, user services, and decision-making processes, ultimately enhancing the library experience for patrons.

Fig. 1.

Data pre-processing in the machine learning process

4. Data Collection and Pre-processing for Machine Learning Models

Data collection and pre-processing are critical stages in building effective machine learning models, as they directly impact the quality and performance of the algorithms. Data collection involves gathering relevant datasets from various sources, while pre-processing encompasses cleaning, transforming, and preparing the data for analysis. These steps are essential to ensure that machine learning models can extract meaningful patterns and insights from the data.

Data collection is the foundation of any machine learning project, as the quality and quantity of the data determine the accuracy and robustness of the models (Heaton, 2016). Datasets can be sourced from internal databases, external repositories, APIs, or web scraping tools, depending on the project requirements. It is crucial to collect diverse and representative data to capture the full spectrum of patterns and variations present in the domain of interest. Additionally, data collection must adhere to ethical guidelines and data privacy regulations to protect sensitive information and ensure transparency in the model development process.

Once the data is collected, pre-processing steps are performed to clean and prepare the data for analysis. According to Brownlee (2020) in a study on data preparation for machine learning: data cleaning, feature selection, and data transforms in Python. Pre-processing involves handling missing values, removing outliers, standardizing or normalizing the features, and encoding categorical variables into numerical format. Pre-processing techniques such as feature scaling, dimensionality reduction, and data augmentation are applied to enhance the quality of the data and improve the performance of machine learning algorithms. Proper pre-processing ensures that the data is in a suitable format for training, validation, and testing the models.

A study by Chicco and Jurman (2020), several challenges are associated with data collection and pre-processing in machine learning and one common challenge is dealing with noisy or incomplete datasets that can introduce bias and errors in the models. Another challenge is managing data from multiple sources with varying formats and structures, which requires careful integration and alignment of the datasets. Moreover, ensuring data consistency, reliability, and relevance throughout the pre-processing stages is crucial to prevent data leakage and ensure the generalizability of the models.

Data collection and pre-processing are foundational steps in building accurate and reliable machine learning models. By collecting diverse and representative data and applying robust pre-processing techniques, researchers and practitioners can optimize the performance and effectiveness of their algorithms. Addressing challenges such as data quality, consistency, and integration is essential to unlock the full potential of machine learning in various domains.

Fig. 2.

Machine Learning Human-provided Data Labelled Model

5. Enhancing Precision through Data Labelling and Annotation

Data labelling and annotation play a crucial role in enhancing the precision and accuracy of machine learning models by providing labelled training data that guides the algorithms in identifying patterns and making predictions (Navab, Hornegger, Wells, & Frangi (Eds.). 2015). Data labelling involves assigning meaningful labels or tags to the data instances, while annotation involves adding descriptive information or metadata to the data to provide context and structure. Alongside carefully annotating and labelling the data, researchers and practitioners improves the quality of the training datasets and optimize the performance of machine learning algorithms.

One of the key benefits of data labelling and annotation according to Kermany (2018) is the creation of a ground truth dataset that serves as a reference for training and evaluating machine learning models. Ground truth data consists of accurately labelled and annotated samples that represent the true underlying patterns in the domain of interest. However, leveraging ground truth data, machine learning models can learn to recognize and generalize from the labelled examples, leading to higher precision and reliability in their predictions (Sokolova & Lapalme, 2009).

Furthermore, data labelling and annotation enable researchers to address class imbalance and bias in the training datasets, which can greatly impact the performance of machine learning models (Japkowicz & Stephen, 2002). By carefully annotating and balancing the distribution of classes in the data, researchers can mitigate bias and ensure that the models learn from diverse and representative examples. This approach enhances the robustness and generalizability of the models, leading to more accurate and unbiased predictions.

Lipton (2018) in a study on the mythos of model interpretability. Data labelling and annotation contribute to the interpretability and explain ability of machine learning models by providing insight into the decision-making process of the algorithms. Annotated data helps researchers understand the features and patterns that the models rely on to make predictions, enabling them to interpret and validate the model’s outputs. This transparency is essential in high-stakes applications such as healthcare, finance, and autonomous driving, where decision-making processes need to be explainable and trustworthy.

Data labelling and annotation are essential processes for enhancing precision and accuracy in machine learning. Alongside creating high-quality labelled datasets, researchers can improve the performance, reliability, and interpretability of machine learning models. Leveraging annotated data for training and evaluation leads to more robust and effective algorithms that can make accurate predictions in various domains.

6. Improving Recommendation Systems in Libraries using Machine Learning

Recommendation systems have become a cornerstone of enhancing user experiences in libraries by providing personalized content suggestions tailored to individual preferences. With the advancement of machine learning algorithms, libraries have the opportunity to revolutionize their recommendation systems, leading to more accurate and insightful recommendations for users.

Ricci et al. (2011) that collaborative filtering is a popular machine learning approach used in recommendation systems to suggest items based on user interactions and similarities among users. However, analysing user-item interactions, collaborative filtering algorithms identify users with similar preferences and recommend items that these users have liked or engaged with. In the context of libraries, collaborative filtering can be leveraged to recommend books, articles, or resources based on the reading habits and preferences of like-minded users, thereby enhancing the relevance and accuracy of recommendations provided.

Lops et al. (2011) report that content-based filtering is another machine learning technique that analyses item attributes and user preferences to generate personalized recommendations. By focusing on the content and features of items in the library collection, content-based filtering algorithms recommend resources that are similar in content or genre to those previously liked by the user. This approach enables libraries to offer targeted and tailored recommendations to users, improving user engagement and satisfaction with the library’s offerings.

Hybrid recommendation systems, which combine collaborative filtering and content-based filtering approaches, offer a comprehensive and diverse set of recommendations to library users (Burke, 2002). Alongside integrating multiple recommendation techniques, hybrid systems can leverage the strengths of each approach to provide more accurate and varied suggestions. In the context of libraries, hybrid recommendation systems can cater to a wide range of user preferences and interests, enhancing the overall recommendation experience for users.

Sarwar et al. (2001) reported that machine learning algorithms enable libraries to incorporate real-time user feedback and behaviour into their recommendation systems, allowing for dynamic adjustments and updates to the recommendations provided. Alongside continuously learning from user interactions, machine learning-based recommendation systems can adapt to evolving user preferences and trends, ensuring that the recommendations remain relevant and up-to-date. This real-time learning mechanism enhances the user experience by delivering timely and personalized recommendations to library users.

Leveraging machine learning techniques in recommendation systems can significantly improve the effectiveness and efficiency of libraries in providing personalized content suggestions to users. Besides employing collaborative filtering, content-based filtering, hybrid methods, and real-time learning mechanisms, libraries can enhance the relevance, accuracy, and user engagement of their recommendation systems. As libraries strive to meet the evolving needs and preferences of their users, incorporating machine learning algorithms into recommendation systems is essential for delivering a more enriching and satisfying user experience.

7. Evaluating the Performance of Machine Learning Models in Library Settings

Machine learning models have gained significant traction in library settings, enabling libraries to enhance their services, optimize resource allocation, and improve user experiences. However, the evaluation of these machine learning models is crucial to ensure their effectiveness and reliability in recommending resources to users.

One key aspect of evaluating machine learning models in library settings is assessing the accuracy of recommendations provided to users. Accuracy measures how well the model predicts user preferences and interests when recommending resources such as books, articles, or multimedia content. In a study on item-based collaborative filtering recommendation algorithms Sarwar, Karypis, Konstan, and Riedl (2001) reported that several metrics are used to evaluate the accuracy of machine learning models, including precision, recall, F1 score, and area under the receiver operating characteristic curve (AUC-ROC). Analysing these metrics, libraries can gauge the performance of their recommendation systems and fine-tune the models to deliver more accurate suggestions to users.

Moreover, evaluating the performance of machine learning models in library settings involves assessing the diversity and serendipity of recommendations. Diversity measures the variety of resources recommended to users, ensuring that the system presents a broad range of options and prevents over-recommending similar items. Serendipity, on the other hand, evaluates the system’s ability to surprise users with unexpected but relevant recommendations, enhancing user engagement and discovery in the library (Hurley et al., 2011). Furthermore, incorporating diversity and serendipity evaluations, libraries can provide users with a more enriching and diverse set of recommendations that cater to different preferences and interests.

Furthermore, evaluating the scalability and efficiency of machine learning models in library settings is essential to ensure their seamless integration and performance in real-world scenarios. Scalability refers to the model’s ability to handle large volumes of data and user interactions efficiently, supporting a growing user base and diverse content collections. Efficiency metrics such as response time, throughput, and resource utilization are crucial for assessing the performance of machine learning models in library settings. Optimizing the scalability and efficiency of recommendation systems enables libraries to deliver timely and responsive recommendations to users, enhancing their overall experience.

Evaluating the performance of machine learning models in library settings is essential for ensuring the effectiveness, accuracy, and user satisfaction of recommendation systems. Alongside assessing metrics such as accuracy, diversity, scalability, and efficiency, libraries can optimize their machine learning models to provide personalized and relevant recommendations to users. As libraries continue to leverage machine learning algorithms to enhance their services and engage users, the evaluation of model performance plays a pivotal role in delivering a seamless and enriching user experience in library settings.

8. Ethical Considerations in Implementing Machine Learning in Libraries

The integration of machine learning technologies in libraries has revolutionized the way libraries provide services, personalize user experiences, and manage resources. However, along with the benefits of machine learning in libraries comes the responsibility to address ethical considerations to ensure fair, transparent, and accountable use of these technologies. Ethical considerations play a crucial role in the successful implementation of machine learning in libraries, safeguarding user privacy, promoting inclusivity, and preventing biases in recommendation systems.

One of the key ethical considerations in implementing machine learning in libraries is the protection of user privacy and data security. A study by Škovrlj (2019) machine learning models analyse user behaviour, preferences, and interactions to generate personalized recommendations, libraries must uphold strict data privacy regulations and ensure that user data is anonymised, encrypted, and securely stored. Libraries should obtain user consent for data collection and processing, clearly communicate how user data will be used, and provide options for users to control their data privacy settings. Besides prioritizing user privacy, libraries can build trust with their users and uphold ethical standards in the implementation of machine learning technologies.

Moreover, ensuring fairness and transparency in machine learning algorithms is essential to mitigate biases and discrimination in library services. Machine learning models learn from historical data, which may contain biases related to race, gender, or socio-economic status. Crawford (2019) urged that libraries must implement fairness and transparency measures, such as bias detection, model explain ability, and algorithmic audits to identify and address discriminatory patterns in recommendation systems. Beyond promoting fairness and transparency in machine learning algorithms, libraries can minimize biases and promote equity in the delivery of services to diverse user groups.

In addition, promoting inclusivity and accessibility in machine learning implementations is critical to ensure that library services cater to all users, including those with diverse backgrounds and needs. Libraries should consider the diverse needs of their user community, including individuals with disabilities, non-native language speakers, or marginalized groups, when designing and deploying machine learning models (Huang & Rust, 2018). Beyond incorporating inclusivity and accessibility features in machine learning systems, libraries can enhance user engagement, improve information access, and foster a welcoming and inclusive environment for all library patrons.

Furthermore, establishing accountability and governance frameworks for machine learning implementations in libraries is essential to reinforce ethical standards, monitor system performance, and address potential harms or misuse of technology. Libraries should establish clear guidelines for data usage, model training, and algorithmic decision-making, as well as mechanisms for auditing, oversight, and redress in case of ethical violations or unintended consequences (Finocchiaro, Maio, Monachou, Patro, Raghavan, Stoica, & Tsirtsis, 2021). Alongside promoting accountability and governance in machine learning implementations, libraries can uphold ethical principles, mitigate risks, and ensure responsible use of technology to benefit their users and communities.

Addressing ethical considerations in implementing machine learning in libraries is paramount to uphold user privacy, fairness, inclusivity, and accountability. However, prioritizing ethical standards in the design, deployment, and governance of machine learning technologies, libraries can build trust with their users, mitigate biases, promote equity, and provide inclusive and accessible services to all patrons. As libraries continue to harness the power of machine learning to enhance their services and engage users, ethical considerations serve as a guiding framework to ensure responsible and ethical use of technology in library settings.

9. Future Trends and Opportunities for Machine Learning in Libraries

Machine learning continues to reshape the landscape of libraries, offering unprecedented opportunities for enhancing user experiences, optimizing resources, and driving innovation in library services. As technology advances and libraries adapt to meet the evolving needs of their users, several future trends and opportunities are emerging in the application of machine learning in library settings. These trends not only present new possibilities for libraries but also pose challenges that require strategic planning and investment in technology and expertise.

One of the key future trends in machine learning for libraries is the personalization of user experiences through recommendation systems. As machine learning algorithms become more sophisticated in analysing user behaviour and preferences, libraries can leverage these algorithms to provide personalized recommendations for content, services, and activities tailored to individual users’ needs (Cordell, 2020). Invariably, implementing recommendation systems powered by machine learning, libraries can enhance user engagement, promote discovery of relevant resources, and improve overall user satisfaction with library services.

Another significant trend is the integration of natural language processing (NLP) capabilities in library systems to enable advanced search and discovery functionalities. According to Kaplan, Downie, and Willer (2016) NLP technologies, such as language understanding and text processing, can be utilized to enhance search interfaces, facilitate semantic search, and extract insights from unstructured library data, such as text archives, digitized collections, and user-generated content. Besides incorporating NLP in library systems, libraries can streamline information retrieval, enable cross-referencing of diverse resources, and support users in exploring complex and specialized topics within library collections.

Furthermore, the adoption of machine learning for metadata enrichment and management presents significant opportunities for improving the discoverability and accessibility of library resources. Gueguen (2019) suggest that machine learning algorithms can automate the process of metadata generation, enriching bibliographic records with descriptive tags, subject classifications, and contextual information to enhance resource categorization and retrieval. However, leveraging machine learning for metadata enrichment, libraries can enhance the organization of their collections, facilitate resource discovery, and improve the usability of library catalogues and digital repositories.

Additionally, the use of predictive analytics and machine learning models for demand forecasting and collection development is a promising trend that can help libraries optimize their resources and collections based on user preferences and trends. Scott and Vogus (2019) report that analysing historical usage data, circulation patterns, and user interactions, libraries can apply predictive analytics to anticipate user needs, identify popular trends, and optimize the acquisition and circulation of library materials. This proactive approach to collection development can help libraries allocate resources effectively, reduce costs, and ensure that collections remain relevant and responsive to user demands.

The future trends and opportunities for machine learning in libraries are vast and promising, offering new possibilities for enhancing user experiences, optimizing resource management, and driving innovation in library services. By embracing these trends and leveraging machine learning technologies, libraries can stay at the forefront of technological advancements, meet the evolving needs of their users, and continue to be vital hubs of knowledge and information in the digital age.

10. Conclusion

The journey towards enhancing the precision of machine learning in the library profession is a transformative endeavour that holds immense potential for revolutionizing how libraries operate, deliver services, and engage with their users. As libraries increasingly embrace and integrate machine learning technologies into their workflows, they are poised to unlock new opportunities for optimizing resource management, enhancing user experiences, and advancing information discovery and access. The power of machine learning algorithms to extract insights from vast datasets, improve search functionalities, and personalize user interactions, libraries can position themselves as dynamic and innovative hubs of knowledge in the digital age. However, this pursuit of precision in machine learning requires diligent attention to data quality, algorithm transparency, and ethical considerations to ensure that the technology serves the ultimate goal of empowering libraries to better serve their communities. As libraries continue to navigate the evolving landscape of technology and information, the quest to enhance the precision of machine learning remains a critical endeavour that will shape the future of the library profession and its capacity to evolve in alignment with the changing needs and expectations of users in the digital era.

References

• Ajani, Y. A., Tella, A., Salawu, K. Y., & Abdullahi, F. (2022). Perspectives of Librarians on Awareness and Readiness of Academic Libraries to Integrate Artificial Intelligence for Library Operations and Services in Nigeria. Internet Reference Services Quarterly, 26(4), 213-230. Advance online publication. [https://doi.org/10.1080/10875301.2022.208619]
• Bi, S., Wang, C., Zhang, J., Huang, W., Wu, B., Gong, Y., & Ni, W. (2022). A survey on artificial intelligence aided internet-of-things technologies in emerging smart libraries. Sensors, 22(8), 2991. [https://doi.org/10.3390/s22082991]
• Brownlee, J. (2020). Data preparation for machine learning: data cleaning, feature selection, and data transforms in Python. Machine Learning Mastery.
• Burke, R. (2002). Hybrid recommender systems: Survey and experiments. User modeling and user-adapted interaction, 12, 331-370. [https://doi.org/10.1023/A:1021240730564]
• Chicco, D., & Jurman, G. (2020). The pitfalls of big data and machine learning in digital healthcare. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 10(6), e1357. [https://doi.org/10.1161/CIRCOUTCOMES.116.003308]
• Cordell, R. (2020). Machine learning and libraries: a report on the state of the field. Library of Congress.
• Crawford, K. (2019). The trouble with bias: A resource list. AI Now Institute.
• Croce, V., Caroti, G., De Luca, L., Jacquot, K., Piemonte, A., & Véron, P. (2021). From the semantic point cloud to heritage-building information modeling: A semiautomatic approach exploiting machine learning. Remote Sensing, 13(3), 461. [https://doi.org/10.3390/rs13030461]
• Das, R. K., & Islam, M. S. U. (2021). Application of artificial intelligence and machine learning in libraries: a systematic review. arXiv preprint arXiv:2112.04573
• de Carvalho, L. F., & da Silva, J. C. (2022). A data-driven Machine Learning approach to creativity and innovation techniques selection in solution development. Knowledge-Based Systems, 257, 109893. [https://doi.org/10.1016/j.knosys.2022.109893]
• Finocchiaro, J., Maio, R., Monachou, F., Patro, G. K., Raghavan, M., Stoica, A. A., & Tsirtsis, S. (2021, March). Bridging machine learning and mechanism design towards algorithmic fairness. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 489-503. [https://doi.org/10.1145/3442188.3445912]
• Georgalis, J., Samaratunge, R., Kimberley, N., & Lu, Y. (2015). Change process characteristics and resistance to organisational change: The role of employee perceptions of justice. Australian Journal of Management, 40(1), 89-113. [https://doi.org/10.1177/0312896214526212]
• Gueguen, G. (2019). Metadata quality at scale: Metadata quality control at the Digital Public Library of America. Journal of Digital Media Management, 7(2), 115-126.
• Heaton, J. B. (2016). “Data preparation for machine learning.” Heaton Research, Inc.
• Huang, M. H., & Rust, R. T. (2018). Artificial intelligence in service. Journal of service research, 21(2), 155-172. [https://doi.org/10.1177/1094670517752459]
• Hurley, N. (2011). The diversity, serendipity, novelty, and coverage of recommendations: A survey. ACM Computing Surveys.
• Japkowicz, N., & Stephen, S. (2002). The class imbalance problem: A systematic study. Intelligent data analysis, 6(5), 429-449. [https://doi.org/10.3233/IDA-2002-6504]
• Kaplan, C., Downie, J. S., & Willer, M. (2016). Natural language processing and the digital humanities. IEEE Intelligent Systems.
• Kermany, D. S. (2018). Deep learning and data labelling for medical applications. IEEE Journal of Biomedical and Health Informatics, 22(3), 1199-1206.
• Lipton, Z. C. (2018). The mythos of model interpretability. Queue, 16(3), 30-57. [https://doi.org/10.1145/3233231]
• Lops, P., De Gemmis, M., & Semeraro, G. (2011). Content-based recommender systems: State of the art and trends. Recommender systems handbook, 73-105. [https://doi.org/10.1007/978-0-387-85820-3_3]
• Navab, N., Hornegger, J., Wells, W. M., & Frangi, A. (Eds.). (2015). Medical Image Computing and Computer-Assisted Intervention-MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III (Vol. 9351). Springer.
• Padilla, T. (2019). Responsible Operations: Data Science, Machine Learning, and AI in Libraries. OCLC Research Position Paper. OCLC Online Computer Library Center, Inc. 6565 Kilgour Place, Dublin, OH 43017. [https://doi.org/10.25333/xk7z-9g97]
• Ricci, F., Rokach, L., & Shapira, B. (2010). Introduction to recommender systems handbook. In Recommender systems handbook, 1-35. Boston, MA: springer US.
• Sarmiento, A. (2021). Effective data annotation for machine learning. IEEE Access, 9, 8561-8577.
• Sarwar, B., Karypis, G., Konstan, J., & Riedl, J. (2001). Item-based collaborative filtering recommendation algorithms. In Proceedings of the 10th international conference on World Wide Web, 285-295. [https://doi.org/10.1145/371920.372071]
• Scott, M. C., & Vogus, B. (2019). Using Data to Support Collection Development and Improve Library Value. Public Services Quarterly, 15(4), 337-343. [https://doi.org/10.1080/15228959.2019.1663337]
• Shah, S. F. A., Iqbal, M., Aziz, Z., Rana, T. A., Khalid, A., Cheah, Y. N., & Arif, M. (2022). The role of machine learning and the internet of things in smart buildings for energy efficiency. Applied Sciences, 12(15), 7882. [https://doi.org/10.3390/app12157882]
• Škovrlj, A. (2019). The Ethics of Artificial Intelligence (Doctoral dissertation, University of Zagreb. Faculty of Economics and Business).
• Sokolova, M., & Lapalme, G. (2009). A systematic analysis of performance measures for classification tasks. Information processing & management, 45(4), 427-437. [https://doi.org/10.1016/j.ipm.2009.03.002]
• Tufail, S., Riggs, H., Tariq, M., & Sarwat, A. I. (2023). Advancements and challenges in machine learning: A comprehensive review of models, libraries, applications, and algorithms. Electronics, 12(8), 1789. [https://doi.org/10.3390/electronics12081789]
• Wheatley, A., & Hervieux, S. (2019). Artificial intelligence in academic libraries: An environmental scan. Information Services & Use, 1, 1-10. [https://doi.org/10.3233/ISU-190065]