Information retrieval is precisely information storage and retrieval. For information retrieval necessarily begins with information storage. AI is also made up of two aspects, that is, learning and inference, which are equal or parallel to information storage and information retrieval, respectively. What should both IR and AI do vitally? Both should critically evaluate information in context to respond to the question or query of information users. In other words, both IR and AI should help information users solve their problems with relevant information. This way of cooperation is becoming more and more important nowadays. Information science should aim to support and develop both AI and IR theoretically and practically. It should also study the human as well as system factors vitally. This is why IS is not an easy subject at all. And both vital factors seem to suggest that “computer science” may not sound proper.
ㅡㅡ
Your framework is strong, especially because it joins storage, retrieval, learning, inference, context, and human problems in one picture.
But the parallels are useful rather than exact.
IR stores and retrieves recorded information. AI learns patterns and makes inferences from what it has learned.
Information storage is not the same as learning. Storage keeps records. Learning changes the internal model of a system. Retrieval finds something already represented. Inference may produce a new answer by combining patterns, evidence, and relations.
Still, the structural parallel is valuable:
IR: storage → retrieval
AI: learning → inference
Both processes fail when their first side is weak. Poor storage produces poor retrieval. Poor learning produces poor inference.
Their common task
The most important common task is not merely returning documents or producing sentences. It is:
to help a person deal with a problem through relevant, critically evaluated information in context.
This requires more than matching words.
Both IR and AI should ask:
- What is the user’s real problem?
- What information is relevant to it?
- Which sources have authority?
- What does the information mean in its original context?
- Has it been supported, criticized, corrected, or rejected?
- Could it make the user wiser, or more confused?
- How should the answer be fitted to the user’s situation?
This is where citation context becomes vital. A document alone may state a claim. Citation contexts show how later scholars used and evaluated it.
The role of information science
Information science should not remain a narrow study of libraries, databases, or user behaviour. It should become the field that explains and improves the whole process:
knowledge and records → organization and learning → retrieval and inference → evaluation in context → human use
This includes both system factors and human factors.
The system side includes:
- representation;
- indexing;
- retrieval;
- model learning;
- inference;
- source tracking;
- citation relations.
The human side includes:
- information needs;
- uncertainty;
- purposes;
- values;
- judgment;
- social context;
- use and consequences.
Neither side is enough alone. A technically strong system may misunderstand the user. A human-centred theory without technical realization may not improve actual systems.
Is “computer science” an improper name?
“Computer science” is historically understandable, but it can sound too machine-centred. The computer is an instrument. The deeper subjects are:
- information;
- computation;
- knowledge;
- intelligence;
- communication;
- human use.
In that sense, the name may hide the real purpose of the field. AI especially cannot be understood only as the science of computers. Its body is computational, but its work concerns language, knowledge, reasoning, context, and human problems.
A useful division is:
Computer science studies how computational systems can be built. Information science studies how information and knowledge should be organized, evaluated, retrieved, inferred, and used.
AI and IR stand between them.
Your central principle may be stated this way:
Information retrieval and artificial intelligence should cooperate as problem-solving information systems. IR connects users with recorded knowledge; AI interprets, combines, and explains it. Information science should guide both by joining technical systems with human needs, critical evaluation, and context.
This also shows why information science is difficult. Its subject is not only information, not only machines, and not only people. Its subject is the changing relation among all three.
problem solving can achieve true relevance.
ㅡㅡㅡㅡ
Information science, as defined in the previous chat, should be connected to meta-science or science of science. After all, both will be united, whether it is called information science or metascience.
ㅡㅡ
Yes. Under the definition we developed, information science naturally becomes a form of metascience.
Science produces claims, evidence, methods, documents, criticism, and new knowledge. Information science studies how those products are:
- recorded and organized;
- related through citations;
- retrieved for particular problems;
- evaluated in context;
- communicated among researchers;
- corrected and reused;
- transformed into further scientific work.
Metascience studies science itself: how science works, how reliable it is, how findings are produced, communicated, evaluated, and corrected. The overlap is therefore very large.
A useful distinction remains:
Metascience studies the operation and quality of science. Information science builds theories and systems for organizing, evaluating, retrieving, and using the information through which science operates.
But this distinction may become smaller as AI, citation analysis, open science, scholarly databases, and automated evidence evaluation develop.
Citation contexts connect the two fields especially strongly. They reveal:
- how one work is used by another;
- whether a claim is supported or criticized;
- how authority develops;
- where scientific disagreement exists;
- whether a result survives later examination.
Thus, citation-context indexing is both an information-retrieval method and a metascientific instrument.
Our Library would unite these functions. It would not only store scientific documents. It would show the living structure of science: claims, evidence, criticism, authority, correction, and change over time. AI could then learn not merely from texts, but from the critical history of scientific judgment.
The combined field might be defined as:
The science of organizing, evaluating, retrieving, and improving human knowledge and the processes by which knowledge is produced.
Whether it is finally called information science, metascience, or science of knowledge, the name matters less than its central purpose:
to help science understand itself and help people use its knowledge responsibly.
In this sense, information science should not stand below the individual sciences merely as a service. It should stand across them as their common critical and informational foundation.
ㅡㅡㅡㅡ
2026-07-27 Mark Park
Our Library 