The Role of Expert Systems in Revolutionizing Reference Services

Walk up to a library reference desk with a quick question-the population of a city, the meaning of a technical term, which database covers a subject-and a skilled librarian will answer in moments. But what happens at 2 a.m., when the desk is empty and the library is closed? This is where expert systems step in. These are computer programs built to mimic the reasoning of a human expert, and for decades librarians have wondered whether such systems could shoulder the steady stream of repetitive, predictable questions that fill a reference desk’s day. Understanding how they work reveals a great deal about both the promise and the limits of automating professional judgement.

What is an expert system? 🔗

An expert system is a branch of artificial intelligence designed to simulate the decision-making ability of a human specialist within a narrow, well-defined field. It does not think the way people do. Instead, it delivers expert-level results by applying stored knowledge to a specific problem. Feed it a question it was built to handle, and it reasons through its rules to produce an answer, a recommendation, or a referral.

The strength of an expert system lies in its focus. A system designed for medical diagnosis cannot help you with engineering, and vice versa. This narrowness is a feature, not a flaw. Within its domain, a well-built system can be consistent, fast, and tireless-qualities that matter enormously when the same kinds of questions arrive again and again.

The core components 🔗

Most expert systems are built from a few essential parts that work together. Knowing these makes the rest of the topic far easier to follow.

Knowledge base: This is the heart of the system, a structured store of facts and rules about the domain. The knowledge held here is usually of two kinds: factual knowledge (data, concepts, and the relationships between them) and procedural knowledge (the if-then rules and strategies an expert uses to solve problems). In a library context, this might include classification standards, the structure of reference sources, and rules about which tool answers which type of question.

Inference engine: This is the reasoning component. It applies the rules in the knowledge base to the input a user provides and draws conclusions, much as a human expert would. It typically uses two strategies: forward chaining, which starts with known facts and works toward a conclusion, and backward chaining, which starts with a goal and works backward to find the facts that support it.

User interface: This is the bridge between the person and the machine. Through dialogue boxes, forms, or simple question prompts, it gathers the user’s query and presents the system’s response in an understandable form.

Many systems add an explanation module, which can justify how it reached an answer, and a knowledge acquisition module for updating the rules over time. The Vidwan shell, developed at the National Centre for Software Technology in Mumbai in 1993, is a notable example of a tool that let developers encode knowledge as IF-THEN rules without building everything from scratch.

Why ready-reference queries are a natural fit 🔗

Reference questions are not all the same. Librarians often sort them into categories, and one of the most common is the ready-reference question-a request for a single fact or short answer that can be settled by consulting one or two standard sources. Who are my elected representatives? What is the date of a historical event? What does a particular term mean? These questions are predictable, recurring, and rule-bound.

That predictability is exactly what an expert system handles well. Because the answer follows a clear path through a limited set of sources, the reasoning can be captured in rules and the work can be automated. The more open-ended and interpretive a question becomes, the harder it is for a system to handle-but the routine factual queries that consume so much of a librarian’s time are well within reach.

Applications in libraries 🔗

The idea is not new. As far back as the 1980s, libraries were experimenting with expert systems for reference work. The National Agricultural Library built a microcomputer-based system called Answerman, designed to help users get answers to simple reference questions by guiding them to the right reference books. A broad survey of the literature found that reference services were among the most active areas of expert-system development in libraries, second only to intelligent front-ends for online searching.

Serving users when the library is closed 🔗

One of the most practical benefits is availability. A reference librarian works fixed hours, but a digital system does not tire and does not close. When students need a quick answer late at night or on a holiday, an expert system can keep the reference function running. It can guide a user to the appropriate database, suggest a suitable reference source, or answer a factual query directly-work that would otherwise have to wait until staff returned.

Easing the burden on staff 🔗

Repetitive questions take a real toll. When the same handful of factual queries arrive dozens of times a day, they consume hours that a librarian could spend on deeper, more interesting work. By taking on these routine transactions, an expert system makes the librarian’s expertise available more consistently and frees professionals to concentrate on complex research consultations, instruction, and the kind of judgement that no rule base can replace. The system handles the predictable; the human handles the nuanced.

Supporting consistency and training 🔗

Because an expert system applies the same rules every time, it answers a given question the same way regardless of who asks or when. This consistency is valuable in large or multi-branch library systems, where service quality can otherwise vary from desk to desk. The same encoded knowledge can also serve as a training aid, helping newer staff learn how an expert would approach a particular type of query.

Advantages and future potential 🔗

The benefits of applying expert systems to reference work cluster around a few clear themes.

Saving time: Routine questions are answered instantly, without a user waiting in a queue and without a librarian repeating the same steps. This efficiency benefits both sides of the desk.

Serving many users at once: A human librarian can help one person at a time. A digital system can respond to many simultaneous queries, scaling in a way that a single desk never can-particularly useful during exam season or other periods of heavy demand.

Handling complexity within a domain: Within their defined field, expert systems can work through layered, rule-based problems and even manage incomplete information, offering structured guidance where a simple keyword search would fail.

Preserving expertise: When an experienced librarian retires, their practical know-how can leave with them. Encoding part of that expertise into a system helps preserve it for future users and staff.

The honest limitations 🔗

It would be misleading to present only the advantages. Early enthusiasm for library expert systems peaked around 1989 to 1990 and then declined sharply. The reasons are instructive. Building a knowledge base is slow and labour-intensive, the rules need constant updating as sources change, and effective reference work involves far more than answering a clearly stated question. A good librarian conducts a reference interview, drawing out what a user actually needs-because people rarely ask for exactly what they want. That interpretive, conversational skill proved very hard to capture in rules.

What this means in the age of AI 🔗

This history matters now more than ever. Today’s AI chatbots are being tested in academic libraries to handle typical information inquiries, and they face many of the same challenges the old expert systems did. Effective reference service requires dialogue, clarification, a deep knowledge base, problem-solving, and knowing when to refer a question elsewhere. One analysis argues that the library reference desk was, in a sense, the original chatbot, and that the lessons from decades of expert-system work-summed up as knowledge, explanation, and flexibility-are exactly what modern AI tools still need to get right.

Seen this way, expert systems are not a relic. They are an early chapter in an ongoing effort to make professional expertise available on demand. The technology has changed, but the goal remains the same: let machines handle the routine so that human professionals can focus on the work that genuinely needs a human mind.

What do you think? If a library system could answer your routine factual questions instantly at any hour, would you trust it for the more complicated research questions too-or are there parts of reference work you would always want a human librarian to handle? And as AI chatbots take on reference tasks, which lesson from the older expert systems-knowledge, explanation, or flexibility-do you think matters most?

References

1. https://www.ebsco.com/research-starters/computer-science/artificial-intelligence-expert-systems
2. https://www.geeksforgeeks.org/artificial-intelligence/what-are-the-different-components-of-an-expert-system/
3. https://www.lenovo.com/us/en/knowledgebase/expert-systems-a-comprehensive-guide/
4. https://www.tutorialspoint.com/artificial_intelligence/artificial_intelligence_expert_systems.htm
5. https://lili.org/forlibs/ce/able/course9/questions-kinds-6.htm
6. https://eric.ed.gov/?id=EJ344246
7. https://www.researchgate.net/publication/332672034_The_application_of_expert_systems_in_libraries_and_information_centres
8. https://www.routledge.com/Expert-Systems-in-Reference-Services/Roysdon-White/p/book/9780367370664
9. https://southernlibrarianship.icaap.org/content/v01n03/olmstadt_w01.html
10. https://en.wikipedia.org/wiki/Reference_desk
11. https://informationmatters.org/2024/11/looking-backwards-to-see-ahead-the-case-of-expert-systems-development-in-libraries/