Loading field map
Information science is built on a persistent tension: how should we organize, retrieve, and understand information, and whose perspective should guide that effort? For over a century, the field has answered this question through a series of competing frameworks, each reacting to the blind spots of its predecessors. Some frameworks have been replaced, others absorbed, and several continue to coexist, dividing the field into distinct research communities with different assumptions about what information is and how it should be studied.
The first systematic framework, Knowledge Organization (1876–Present), emerged from the practical need to arrange physical collections. Melvil Dewey's decimal classification (1876) gave libraries a standard shelf-location system, but the framework's deeper commitment was to the idea that subject categories could be universal and that a well-designed classification could make any document findable. Knowledge Organization remains active today, though its focus has shifted from shelf arrangement to conceptual structures such as thesauri, ontologies, and linked data vocabularies.
The Documentation Movement (1895–1950) expanded Knowledge Organization's scope. Paul Otlet and Henri La Fontaine argued that classification should cover not just books but all recorded information—articles, patents, images, even ephemera. Otlet's Universal Decimal Classification (UDC) attempted to create a single indexing language for the world's knowledge, and his vision of a networked "réseau" anticipated hypertext and the web. The Documentation Movement declined in the mid-twentieth century partly because its universalist ambition outpaced the available technology: manual indexing of all recorded knowledge was simply too slow and expensive. Yet its core insight—that information science should treat any recorded artifact as a potential object of organization—survived in later frameworks.
After World War II, information science took an engineering turn. The Physical Paradigm (1948–1970) treated information retrieval as a problem of matching query terms to document terms using mechanical or electronic systems. Its methods were borrowed from signal processing and communication theory: term-frequency counts, Boolean logic, and early statistical models. The Physical Paradigm narrowed the field's focus from organizing knowledge to retrieving documents, and it assumed that relevance was a property of the document-query match rather than of the user's situation.
The Cranfield Information Retrieval Paradigm (1957–Present) formalized this system-centric logic. Cyril Cleverdon's experiments at Cranfield in the late 1950s and early 1960s introduced a standardized methodology: a fixed test collection of documents, a set of queries, and human relevance judgments. Precision and recall became the gold-standard metrics. The Cranfield Paradigm did not replace the Physical Paradigm so much as give it a rigorous experimental infrastructure. Today, the Cranfield evaluation model is the backbone of the Text REtrieval Conference (TREC), which has run since 1992. It remains the dominant framework for evaluating retrieval algorithms, even as later frameworks have questioned whether its laboratory conditions capture real-world information needs.
By the late 1960s, researchers began to argue that system metrics missed the human experience of seeking information. Information Behavior (1968–Present) shifted the unit of analysis from the document to the person. Early studies by Thomas D. Wilson and others mapped how people encounter, seek, and use information in everyday life and professional settings. Information Behavior is a broad, empirically driven framework that coordinates research on information needs, information seeking, and information use, often using qualitative methods such as interviews, diaries, and ethnographic observation.
The Cognitive Paradigm (1970–Present) gave the user turn a theoretical engine. Drawing on cognitive psychology, it argued that information retrieval should be understood as a process of matching the user's mental model—their "Anomalous State of Knowledge" (ASK), in Nicholas Belkin's influential formulation—with the knowledge structures embedded in documents. The Cognitive Paradigm explicitly reacted against the Physical Paradigm: relevance was not a property of the document but of the relationship between the document and the user's cognitive state. By the late 1980s, however, the Cognitive Paradigm turned its critique inward. Researchers such as Birger Hjørland and Peter Ingwersen recognized that treating the individual user's mind as the ultimate ground of relevance ignored the social and cultural contexts that shape what people know and what they need. This internal self-critique (1988–1995) opened the door to frameworks that located meaning not in individual cognition but in communities, disciplines, and institutions.
Three frameworks emerged in parallel to address the social dimensions that the Cognitive Paradigm had neglected. Informetrics (1979–Present) took a quantitative, macro-level approach. It applied statistical laws—Lotka's law of author productivity, Zipf's law of word frequency, Bradford's law of journal scattering—to measure patterns in publication, citation, and usage. The h-index, introduced in 2005, became a widely used (and controversial) metric for individual research impact. Informetrics coexists with the Cranfield Paradigm in the evaluation community: where Cranfield measures system performance, Informetrics measures the structure of scholarly communication.
The Domain-Analytic Approach (1995–Present), developed by Birger Hjørland, directly challenged the Cognitive Paradigm's individualism. Hjørland argued that information science should analyze the epistemological and pragmatic norms of specific knowledge domains—disciplines, professions, or discourse communities—rather than the mental states of individual users. A domain-analytic study might examine how physicists classify their literature differently from historians, or how the indexing of medical literature reflects the epistemological commitments of evidence-based medicine. The Domain-Analytic Approach absorbs the Cognitive Paradigm's interest in meaning but relocates meaning from the individual mind to the collective practices of a domain.
Social Informatics (1998–Present), associated with Rob Kling, focused on the socio-technical configuration of information systems. It rejected technological determinism—the idea that a system's design alone determines its effects—and instead examined how organizational context, work practices, and power relations shape the design, implementation, and use of information technology. Social Informatics overlaps with the Domain-Analytic Approach in its attention to context, but it is more concerned with the interplay between technology and social structure than with the epistemological norms of disciplines.
Critical Information Studies (2010–Present) extends the social turn into explicit questions of power, justice, and marginalization. Drawing on feminist theory, postcolonial theory, critical race theory, and queer theory, CIS examines how information systems and practices produce, reinforce, or challenge social inequalities. Where Social Informatics might study how a hospital's electronic health record system affects clinical workflows, CIS would ask how that same system encodes racial or gender biases in its classification categories or access rules. CIS is the youngest framework in the timeline, and it remains in active dialogue with the Domain-Analytic Approach and Social Informatics, sharing their commitment to context while foregrounding normative critique.
Eight of the ten frameworks remain active, and they divide the field into overlapping but distinct research communities. The Cranfield Paradigm dominates information retrieval evaluation, where precision, recall, and TREC-style test collections are the standard tools. Information Behavior and the Cognitive Paradigm lead user studies, with Information Behavior providing the empirical methods and the Cognitive Paradigm supplying the theoretical models of information seeking. Knowledge Organization continues as the home for classification theory, metadata, and ontology design, often in collaboration with the Domain-Analytic Approach, which provides a rationale for domain-specific classification. Informetrics is the quantitative arm of the field, used in research evaluation, bibliometric mapping, and science policy. Social Informatics and Critical Information Studies occupy the socio-technical and critical edges of the field, often in schools of information and communication.
What do these frameworks agree on? Nearly all of them reject the idea that information is a neutral, objective substance waiting to be retrieved. Even the Cranfield Paradigm, the most system-centric, acknowledges that relevance is a human judgment. There is broad consensus that context matters—whether that context is the user's cognitive state, the domain's epistemological norms, or the organization's power structure.
Where they disagree is on the unit of analysis and the epistemological stance. The Cranfield Paradigm treats the system as the primary object of study; Information Behavior and the Cognitive Paradigm treat the individual user; the Domain-Analytic Approach treats the domain; Social Informatics treats the socio-technical configuration; Critical Information Studies treats power and marginalization. These are not competing claims to the same truth but different lenses on a complex phenomenon. The field's vitality comes from the friction between them: the Cranfield Paradigm's precision challenges the qualitative richness of Information Behavior; the Domain-Analytic Approach's collectivism challenges the Cognitive Paradigm's individualism; Critical Information Studies' normative stance challenges the descriptive neutrality of Informetrics. No single framework has absorbed the others, and the history of information science suggests that none will.