Relationships: a primer

Mark R. Lindner

590RO, Spring 2007

Bean and Green

Bean, Carol A. and Rebecca Green, Eds. Relationships in the Organization of Knowledge. Information Science and Knowledge Management, v. 2. Dordrecht: Kluwer Academic Publishers, 2001.

Green, Rebecca, Carol A. Bean and Sung Hyon Myaeng, Eds. The Semantics of Relationships: An Interdisciplinary Perspective. Information Science and Knowledge Management, v. 3. Dordrecht: Kluwer Academic Publishers, 2002.

Khoo, Christopher S. G. and Jin-Cheon Na. "Semantic Relations in Information Science." In Blaise Cronin, Ed. Annual Review of Information Science and Technology 40, 2006: 157-228.

Genesis was participation of the editors in an ACM/SIGIR workshop, "Beyond Word Relations." This shared participation led to the idea of an edited volume, which eventually became two (GBM 2002, vii).

The first volume "examines the role of relationships in knowledge organization theory and practice, with emphasis given to thesaural relationships and integration across systems, langauges, cultures, and disciplines. The second volumes examines relationships in a broader array of contexts. The two volumes should be seen as companions, each informing the other" (GBM 2002, vii).

What got me interested in this in the 1st place?

In mid-December 2006 I was working on a paper for Carole Palmer's 590TR Information Transfer and Collaboration in Science seminar. My paper was to be a "representative literature review on the topic of mapping different thesauri and the uses of such for the organization of information to meet the needs of interdisciplinary scientists." As such, some article I read—now lost to me—pointed me to a chapter in ARIST 40 (2006) by Chris Khoo and Jin-Cheon Na on "Semantic relations in information science." Many of the articles I was reading on mapping thesauri raised the issue of whether or not inter-concept relationships within a single thesaurus could truly carry over into multiply-mapped thesauri, especially in the context of multilingual thesauri. I was considering my problem of mapping across scientific domains to be very similar to "true" multilingual mapping, thus, I decided this might be a highly relevant piece to read. I ended up finding it fascinating!

In a small section on the relevance relationship the author's wrote:

Finally, an important type of semantic relation in information science is the relevance relation—the relevance of a document to a query or to the information need of a user. Researchers have identified many factors, in addition to topical relevance, that affect a user's judgment of the relevance of a document (cites). Green (2001) suggested that there may be several types of semantic relations underlying these factors, which have not been studied in depth. Green and Bean (1995) and Bean and Green (2001) have explored some of the relations underlying topical relevance (Khoo and Na 2006, 174).

Hmmm? I'm looking at mapping across scientific disciplines; it seems relevance should be highly important. I also am aware of the seemingly unorthodox associations I make between literatures and am, thus, well aware that relevance does not equate to topicality. The authors also cited assorted combinations of Bean & Green, and many of the articles in these 2 edited compilations. I decided to trace this literature and it has been a pleasant uphill climb ever since!

Editors

Carol Bean ???

  • Division of Biomedical Technology and Research Resources, National Center for Research Resources, National Institutes of Health, Bethesda, MD (2004 JASIST article).
  • Extramural Programs, National Library of Medicine, Bethesda, MD (2002 BGM)
  • School of Information Sciences, University of Tennessee, Knoxville, TN (2001 B&G)

Rebecca Green

Dr. Rebecca Green has recently joined OCLC to become the assistant editor, Dewey Decimal Classification, based out of the Dewey Editorial Office at the Library of Congress. Prior to this she was an Associate Professor at the University of Maryland College of Information Studies. Her research interests include: Information storage and retrieval, classification theory, database design, cognitive linguistics.

Degrees:

  • B.A., Music, Harvard University
  • M.L.S., Library & Information Studies, University of Maryland
  • M.A., Linguistics, University of California
  • Ph.D., Library & Information Studies, University of Maryland
  • Ph.D., Computer science, University of Maryland

Purpose and Structure of B&G

Addresses relationships "involved in the organization of recorded knowledge" (vii).

Addresses relationships "involved in the organization of recorded knowledge" (viii).

Organized into 2 parts:

  • ROK: Theoretical Background
    • Overview; Bibliographic; Thesaural; Standards (thesauri/indexes); Relationship compatibility in multilingual contexts; Vocabulary integration (UMLS); Cultural dependence of relationship systems; Relevance
  • ROK: Systems
    • LCSH; AAT; MeSH; OMIS and explicit associative ("lateral") relationships; Colon Classification; DDC

Bean & Green's Taxonomy of Relationships

Not mutually exclusive (vii)

[See slide]

Green: An Overview

Just a heads up to where we�re going.

Overview: Properties of relationships 1

Def: "In the broadest sense, a relationship is an association between two or more entities or between two or more classes of entities. To specify a relationship, we must be able, first, to designate all the parties bound by the relationship and, second to specify the nature of the relationship" (3).

May distinguish 2 types: abstract and concrete. Abstract relationships express association between entity classes, while concrete relationships express associations between specific entities (3). [Examples on slide]

Abstract relationships are used in data modeling, while concrete relationships are recorded in an actual database. Data modeling of bibliographic entities, e.g., FRBR, uses abstract relationships, while catalogs (generally) record concrete relationships (3-4).

Arity: Number of entities bound by a relationship. Two entities or entity classes would be a binary relationship, three a ternary, are more generically we can say an n-ary relationship. Our two examples on the slide are binary, while Supplier < supplies > Part < for > Project is a ternary relationship.

Most relationships are asymmetric. It is not necessarily the case that if John likes Mary, that Mary likes John. And if A is bigger than B, then B is most certainly not bigger than A. < is a cousin of >, < is a sibling of >, and < is a spouse of > are examples of symmetric relationships. Synonymy is theoretically a symmetric relationship, although many would argue that there are no, or few, truly synonymous relationships.

Asymmetry implies "that it must be possible to specify of a given entity (class) which participant it is in the relationship" (4).

Overview: Properties of relationships 2

Nature of the relationship

"Specifying the nature of a relationship is accompanied by its own array of properties..." (4).

Cardinality: one-to-one, one-to-many (or many-to-one), and many-to-many

  • One-to-one: Husband < is currently legally married in a monogamous society to > Wife
  • One-to-many: Man < is biological father of > Child
  • Many-to-many: Person < is parent of > Child

Transitivity: If A < is related in a certain way to > B and B < is related in that same way to > C, and A < is related in that same way to > C, then the relationship is transitive. "Hierarchical inheritance is a type of transivity: If a Faceted thesaurus < is a > Thesaurus and a Thesaurus < is a(n) > Index language, then a Faceted thesaurus < is a(n) > Index language" (5).

"The single most important aspect of specifying the nature of the relationship that holds between entities or entity classes is identifying the semantics of the relationship. This may be done with various degrees of implicitness/explicitness" (5).

  • Count on the enumeration of participants or participant types to imply an underlying relationship. For example see also or RT references in thesauri.
  • A relationship type between entity classes holds by convention. For example Space and Time in Ranganation's PMEST formula. Space and Time are understood to situate the topic as occuring in the specificed Space and Time and not to have been written in that Space and Time.
  • Simply name the relationship type that holds, as I have done so far. Assumes the user is familiar with the relationship type through their own personal experience and use of natural language.
  • Most explicit uses a formal language to set forth the semantics of the relationship type (5).

Considerable effort has been expended on identifying a comprehsive inventory of relationship types, but there is little consensus. There is some consensus on, say, hierarchical relationships, but little on associative relationships (5).

The Open (non-enumerable) vs. Closed (enumerable) class distinction is appropriate here. For instance, nouns, verbs, adjectives, and adverbs are generally open classes, while pronouns, prepositions, and conjunctions are closed classes. Relationships are generally expressed via use of verbs, prepositions and conjunctions and, thus, there are going to be open and closed classes of relationships (6).

Paradigmatic vs. Syntagmatic: This distinction is traced to de Saussure, who called paradigmatic relations associative relations (Khoo & Na, 166, 208).

"Paradigmatic relations are relations between pairs of words or phrases that can occur in the same position in the same sentence (Asher, 1994, v. 10, p. 5153). The words often are instances of the same part of speech, belong to the same semantic class, and are to some extent grammatically substitutable. Examples include IS-A (broader-narrower), part-whole, and synonym relations. These relations tend to be part of our semantic memory, and are typically used in thesauri. Lancaster (1986) chararcterized paradigmatic relations as a priori or permanent relations.

Syntagmatic relations refer to relations between words that co-occur (often in close syntactic positions) in the same sentence or text (Asher, 1994, v. 10, p. 5178). It is a linear or sequence relation that is synthesized or expressed between two words or phrases when we construct a sentence. The relations are governed partly by the syntactic and grammatical rules of a language. Lancaster (1986) characterized syntagmatic relations as a posteriori or transient relations. Green (2001) suggested that paradigmatic relations are a closed, enumerable class of relations, whereas syntagmatic relations are an open class that cannot be fully enumerated,..." (Khoo & Na, 166)

As straightforward as all this sounds, the distinction between paradigmatic and syntgmatic relationships is fuzzy (Khoo & Na, 166).

Green has also written fairly extensively on syntagmatic relationships and argues for their inclusion in information retrieval. Some of those articles will be listed in my annotated bibliography project.

Overview: Relationships in KO

  • Major entity types
  • Bean & Green's Taxonomy of Relationships
    • Bibliographic relationships between units of recorded knowledge
    • Intratextual and intertextual relationships, including those based on text structure, citation relationships, and hypertext links
    • Subject relationships in thesauri and other classificatory structures
    • Relevance relationships

Not mutually exclusive (vii)

Major entity types in the organization of knowledge:

  • Bodies: authors, translators, editors, publishers,...
  • Bibliographic units, both intellectual—texts,...; and physical—books, serials, Web pages,...
  • Subjects, concepts, words and knowledge (both public [in documents] and private [in minds].
  • Users and their information needs (7).

Bean & Green's Taxonomy of Relationships

  • Bibliographic relationships between units of recorded knowledge
  • Intratextual and intertextual relationships, including those based on text structure, citation relationships, and hypertext links
  • Subject relationships in thesauri and other classificatory structures
  • Relevance relationships

Notice that these are not mutually exclusive (vii)

Overview: Bibliographic relationships

  • Def: "include all relationships involved in the descriptive cataloging of bibliographic units, whether considered as physical or material units, on the one hand, or as intellectual units, on the other hand" (7).
  • Abstract level
  • Concrete level
  • Implementation

Broadly defined as "includ[ing] all relationships involved in the descriptive cataloging of bibliographic units, whether considered as physical or material units, on the one hand, or as intellectual units, on the other hand" (7).

Abstract level: 2 primary questions need to be addressed: "What classes of bibliographic entities need to be recognized?" And, "How are these general classes related to each other?" (7).

Several answers have been proposed to these questions. The primary one on the table currently is the IFLA FRBR model, which recognizes Works, Expressions, Manifestations and Items. The relationships between them are that a Work is realized through an Expression which is embodied in a Manifestation which is exemplified by an Item (FRBR, 13).

Concrete: Again, much work has been done here. The primary scholarly investigations are the dissertations of Tillett (1987), Smiraglia (1992), and Vellucci (1997). Tillett's work is the broadest and has been published in various venues, including as the 2nd chapter in the book at hand. A 1998 article by Vellucci gives a good introduction to these important empirical investigations. In fact, Tillett in her article in B&G states "For a complete historical background on research in bibliographical relationships see Vellucci (1998)" (Tillett 2001, 33).

Implementation: Several people have given overviews or commented on various ways in which the implementation, or lack thereof, of relationships between bibliographic units has been undertaken. for instance, see Vellucci 1998.

Overview: Intra- & intratextual relationships

3 types (recursive)

  1. Internal structure of an Expression
  2. Citation
  3. Hypertext linking
    • 2 & 3 are assymmetric (directionality counts)

3 types:

Internal structure: Texts have internal structure, being composed of chapters, sections, subsections and so on. These units exhibit semantic unity. "Text structures operate on all levels of the breakdown, from macrostructures to microstructures" (9). There exists a vast text structure literature (9).

Citation: There is, also, a vast citation literature. For a good intro to this literature, among other ideas, consider taking Carole Palmer's 590TR, which I believe is being offered again this Fall.

Green writes, "Of relevance to knowledge organization is whether citation relationships between Manifestations are paralleled by semantic relationships between their subject matter. If so, we can use citation relationships to substitute for or to supplement more conventional means (e.g., subject indexing, keyword searches) in subject searching" (9). A study by Green (2000) investigated the practices of humanities scholars who prefer citation tracing to searching, and demonstrated the complementarity of the two approaches. This is one of my favorite articles in that it is interesting and well written and it justifies my primary means of finding sources (9).

There have been several studies investigating this relationship, but all have suffered from the limitation of using co-occurrence of specific words and phrases and thus missed the impact of synonymy and hyponymy on the semantic relationships between cited and citing documents (9).

Hypertext linking: While conceptually similar to print-based citation, hypertext links can make direct reference to other documents or document fragments. This ability raises 2 questions, "Are typed links—that is, links in which the semantic relationship between source and destination is made explicit—beneficial to hypertext users?" and, "If so, can link types be generated automatically?" (9-10).

Work has been done to inventory types of hypertext link relationships. And while there are certainly many types, almost none are unique to this context (10).

Overview: Subject relationships

Most important variable in IR

  • Between subjects vs. between concepts
  • Inventory
  • Most typical use
  • 2nd major use
  • 3rd promising use

Arguably, subject is the "single most important variable in information retreival .... Unfortunately, searching by subject is inherently difficult" (10). There are several reasons for this:

  1. Documents are rarely about a single subject.
  2. Users routinely fail to state the exact subject or subjects that can meet their information needs.
  3. A seemingly infinite set of subjects (10).

It is the perceived interrelationship or web of subject relationships that allows for navigating between them (10-11). Of course, it is important to recognize that we are really discussing relationships between concepts, since concepts can be expressed via words or notational systems.

Several folks have undertaken a comprehensive inventory of relationships between subjects. Perhaps the most famous—and widely reprinted— is Perreault's 1965 article, "Categories and relators: a new schema," which identifies and classifies 120 types of intersubject relationships. See this chapter for many more articles addressing types, characterizations, and the expression of relationships amongst subjects.

Primary use of subject relationships is determining which subjects are the most appropriate to use for searching (11). A second major use is to use them "to form more complex (and more precise) subject terms" (12) for indexing and searching, e.g., see Perreault. A third possibility is to use subject relationships in literature-based knowledge discovery. See the work of Swanson (and others) on this topic (12).

Overview: Relevance relationships

  • Characterized:
    • as "a measure of effectiveness between a source and a destination in a communication process. A measure is a relation. Relevance is also a relation" (Saracevic, 1976).
    • by Lancaster and Warner (1993), "in terms of relationships between a document and a user, a request, and/or a need" (Green, 12).
  • Which aspects matter?
    • Topicality
    • Other criteria

Relevance relationships have been characterized by Saracevic as "a measure of effectiveness between a source and a destination in a communication process. A measure is a relation. Relevance is also a relation" (Saracevic, 1976). Lancaster and Warner (1993) characterized relevance "in terms of relationships between a document and a user, a request, and/or a need" (Green, 12).

Of specific import, Green writes, is "the specification of which aspects of the user and the user's need, on the one hand, and which aspects of sources, on the other hand, participate in establishing relevance relationships and what the exact nature of these relationships may be" (12).

Topicality may be the most important aspect of relevance, but many other things can make a big difference. Some of these are:

  • those pertaining to the information content of the documents, e.g., scope, validity, clarity and recency.
  • user's previous experience and background, e.g, ability to understand, and content novelty.
  • user's beliefs and preferences, e.g, subjective validity and affectiveness.
  • other information and sources within the information environment, e.g., consensus and external verification.
  • the sources of the documents, e.g., source quality and source reputation.
  • document as a physical entity, e.g., obtainability and cost.
  • user's situation, e.g., time constraints and relationship with author.

"Each of these factors may be involved in unique types of associations relating relevant material to the user need. However, the relational aspects of these associations have not been addressed in depth" (13). While there is a chapter in this book by Bean and Green on relevance relationships, I recommend the 2 1995 JASIS articles by the same authors on which it is based.

Many have equated relatedness of topic with sameness. This is contra to the everyday notion of relevance and is amply demonstrated by examples from Harter: "That there was a drought in South Dakota in 1985 was relevant to my vacation plans there that year. [pause] Developments in computer technology are relevant to the future careers of students enrolled in schools of library and information science" (Green, 13). The two concepts represented in each statement are clearly conceptually related without being on the same topic at all.

It seems that there exists a vast complexity and range of relationships involved in relevance (13).

Overview: Conclusion

"On the one hand, the relationships involved in knowledge organization are both numerous and often complex.

On the other hand, their very magnitude and complexity militate against their consistent use by information professionals, much less by end users" (13).

Green writes: "On the one hand, the relationships involved in knowledge organization are both numerous and often complex. On the other hand, their very magnitude andd complexity militate against their consistent use by information professionals, much less by end users. Even when subject relationships have been kept to the bare minimum—equivalence, hierarchical, associative—there is sometimes lack of consensus on how to treat a specific relationship, and end users often don't understand what is being communicated by the standard relationship notations. At the same time, end users typically understand the relationships of concern to their information situations intuitively" (13).

Thus, the question of whether a broader and/or deeper understanding and emphasis on relationships in knowledge organization would be beneficial or worse arises (14). Green argues, and I agree, that the increasing explosion of information in various forms and its increasing management by automated means requires a better understanding of the expression and manipulation of relationships in KO (14). Better understanding may well lead to greater precision in topical relationships, and better precision via ability to further specify other user-centered relevance relationships. "Conversely, if we learn how to reason on these relationships, the higher quality may also come in the identification of relevant sources that would otherwise not have been retrieved at all, for example, if we can retrieve sources that are relevant by analogy" (14).

Remainder of Part I

  • Tillett on Bibliographic relationships
  • Dextre Clark on Thesaural relationships
  • Milstead on Standards (thesauri/indexes)
  • Hudon on Relationships in multilingual thesauri
  • Bodenreider & Bean on Vocabulary integration (UMLS)
  • Beghtol on Cultural dependence of relationship systems
  • Bean & Green on Relevance relationships
[See slide]

Tillett - Bibliographic relationships

Reviews and expands on her 1987 dissertation work

  • Equivalence relationships
  • Derivative relationships
  • Descriptive relationships
  • Whole-part relationships
  • Accompanying relationships
  • Sequential relationships
  • Shared characteristic relationships

Reviews and expands on her 1987 dissertation work, particularly in the context of FRBR

  • Equivalence relationships: holds between exact copies of the same manifestation of a work or between an original item and reproductions of it.
  • Derivative relationships: holds between a work and modifications based on it: variations or versions (editions, revisions, translations,...); adaptation or modifications that become new works; changes of genre; new works based on the style or thematic content such as paraphrases and parodies.
  • Descriptive relationships: holds between an entity and a description, criticsim, or review....
  • Whole-part relationships: holds between an entity and a component part, such as between an anthology and an individual selection from it.
  • Accompanying relationships: holds between an entity and its accompanying materials. Sometimes one entity is predominant (text and its supplements, or an entity and its concordances or indexes), and sometimes equal as in a kit.
  • Sequential relationships: holds between entities that continue or precede each other; successive titles of a serial, sequels of a monograph...
  • Shared characteristic relationships: hold between entities that are not otherwise related but coincidentally have a common author, title, subject, or other characteristic used as an access point, such as shared language, date of publication, or country of publication (19-20).

Calls for an expanded and more explicit use of bibliographic relationships via rule changes and use of technological means.

Thesauri times three, plus one

  • Dextre Clark "Thesaural relationships"
  • Milstead "Standards for relationships between subject indexing terms"
  • Hudon "Relationships in multilingual thesauri"
  • Bodenreider & Bean "Relationships among knowledge structures: Vocabulary integration within a subject domain" (UMLS)

Dextre Clark "Thesaural relationships"

Provides a good overview of the standard thesaural relationships of equivalence, hierarchy and association. Points out that the "assumptions, techniques, and conventions" of thesaural relationships were developed in the 1960s and codified into standards in the 1970s (50). While these are still generally valid in many applications, they "begin to break down where processes are automated and/or multiple databases are to be searched simultaneously" (50). Thus, calls for greater relational explication for automated uses.

Milstead "Standards for relationships between subject indexing terms"

Provides a good overview of thesaural standards, although the article is now based on outdated standards. Calls for research into display issues. Questions the value of distinguishing relationship types, but calls for further research. "...,we should be thinking both about what kinds of relationships can be useful at all, and to whom or what they will be useful" (64).

Hudon "Relationships in multilingual thesauri"

Provides a succinct introduction into issues of thesaural relationships in a multilingual context. Asks two primary questions, "Are all types of thesaural relations transferable from one language to another? and Are two members of a valid relation in a source language always the same in the target language(s)?" (67).

Bodenreider & Bean "Relationships among knowledge structures: Vocabulary integration within a subject domain" (UMLS)

The Unified Medical Language System of the NLM "provides a common interface to about 40 existing medical vocabularies..." (82). Concentrates on synonymy, hierarchical relationships and explicitly mapped relationships within the UMLS.

Beghtol: "Rethink all aspects"

"Relationships in classificatory structure and meaning"

  • "Every classification system is a theoretical construct imposed on "reality"" (99).
  • Relevant research
  • Structure and meaning in bibliographic CS
  • "Cultural warrant"
  • Parts & wholes: Universal?

Beghtol's premise "is that changing knowledge structures and the increased globalization of information exchange require rethinking all aspects of bibliographic classification systems, including the kinds of relationships we habitually include in the systems. ... Its general purpose is to raise questions, identify issues, and suggest potentially useful research areas" (99).

She states that "Every classification system is a theoretical construct imposed on "reality"" (99).

Provides an overview of relevant research, looks at structure and meaning in bibliographic classification systems, and examines the idea of "cultural warrant.

Parts & wholes, while fairly well established as cultural universals are still "culturally" determined. Beghtol compares and contrasts both halves of the whole-part relationship, taxonomic subdivision (kinds of) and partonomic subdivision (parts of). The interesting thing, as she points out, is that "the formation of "kind of" and "part of" hierarchies is not neutral but is culturally determined" (107). Both different cultures and different purposes will lead to different subdivisions.

Calls for more research into relationships used in bibliographic classifications.

B&G: Relevance relationships

Green, Rebecca. (1995). "Topical relevance relationships: Why topic matching fails." Journal of the American Society for Information Science 46: 646-653.

Green, Rebecca and Carol A. Bean. (1995). "Topical relevance relationships: An exploratory study and preliminary typology." Journal of the American Society for Information Science 46: 654-662.

While it is a good synthesis of the 2 1995 JASIS articles, I recommend those first. I will also not dally here as I covered this topic as in depth as I can afford in this presentation earlier. I do recommend either this article or the 2 JASIS articles, though, as this is a fascinating topic. I only ask you to think of your own citation practices and muse on what might, in fact, count as relevance.

Part II - Systems

  • El-Hoshy on Relationships in LCSH
  • Molholt on Relationships in AAT
  • Nelson, et. al. on Relationships in MeSH
  • Neelameghan on OMIS and explicit associative ("lateral") relationships
  • Satija on Relationships in CC
  • Mitchell on Relationships in DDC

Part II: Relationships in the organization of knowledge: Systems "address the relational structure of several specific thesauri and classification schemes. El-Hoshy thoroughly explores the expression of relationships in the Library of Congress Subject Headings (LCSH), while Molholt and Nelson, Johnston, and Humphreys do the same for the Art & Architecture Thesaurus (AAT) and Medical Subject Headings (MeSH) respectively. Each of these systems is highly conscious of the importance of relationships and deals with them in somewhat unique ways. Neelameghan introduces a new system, OMIS, which cuts across cultures and religious traditions and uses an extensive set of explicit associative ("lateral") associations. Satija explores the role of relationships in the Colon Classification (CC), while Mitchell addresses the relational structure of the Dewey Decimal Classification (DDC)" (viii).

This part is fascinating in that the articles show the extent, and/or lack thereof, of the implementation of relationships within various thesauri and classification systems. Scope notes can also serve as identifiers of relationships in many of these systems. The AAT is particularly interesting for the process used to create it, and for the fact that in many ways it does its own thing. MeSH is as always fascinating. I recommend these articles for general reading, but more importantly when needed.

GBM: Semantics of Relationships

Green, Bean & Myaeng (2001) The Semantics of Relationships: An Interdisciplinary Perspective.

  1. Types of Relationships
  2. Relationships in Knowledge Representation and Reasoning
  3. Applications of Relationships

The 2nd volume of the pair is edited by Green, Bean and Sung Hyon Myaeng and is titled, The Semantics of Relationships: An Interdisciplinary Perspective. It is divided into 3 parts:

  1. Types of Relationships
  2. Relationships in Knowledge Representation and Reasoning
  3. Applications of Relationships

Having read all the Part I, most of Part II, but none of Part III, I can state that this is also a fascinating volume. I have been intrigued by articles entitled, "Hyponymy and its varieties," "On the Semantics of Troponymy," "Meronymic Relationships: From Classical Mereology to Complex Part-Whole Relations," "The Many Facets of the Cause-Effect Relation," "Internally-Structured Conceptual Models in Cognitive Semantics," and even "Identity and Subsumption." I am looking forward to the rest and the re-reading of many of them.

Khoo & Na "Sem rels in info science."

  • Overview
  • What are Semantic Relations
  • Types of Semantic Relations
  • Selected Semantic Relations
  • Semantic Relations in Knowledge Structures
  • Automatic Identification of Semantic Relations
  • Semantic Relations in Information Retrieval
  • Conclusion
  • References (approx. 20 pages)
  • Overview
  • What are Semantic Relations
    • Semantic Relations in Language and Logic
    • The Psychological Reality of Semantic Relations
    • Semantic Relations in Semantic Memory
  • Types of Semantic Relations
    • Overview
    • Lexical-Semantic Relations
    • Case Relations
    • Relations Between Larger Text Segments
  • Selected Semantic Relations
    • Hyponym-Hyperonym Relation
    • Troponymy Relation
    • Meronym-Holonym Relation
    • Synonymy
    • Antonymy
    • Cause-Effect Relation
  • Semantic Relations in Knowledge Structures
    • Semantic Relations in Thesauri
    • Semantic Relations in Indexing Languages
    • Semantic Relations in Ontologies
  • Automatic Identification of Semantic Relations
    • Overview
    • Automatic Identification of Semantic Relations Using Pattern Matching
    • Automatic Construction of Extraction Patterns
    • Text Mining for Semantic Relations
    • Automatic Construction of Case Frames
  • Semantic Relations in Information Retrieval
    • Overview
    • Semantic Relations in Query Expansion
      • Query Expansion Using Term Association
      • Query Expansion Using Lexical-Semantic Relations
    • Relation Matching for Precision Enhancement
    • Question-Answering with Full-Text Documents
    • Semantic Relations in Automatic Text Summarization
  • Conclusion
  • References (approx. 20 pages)

Relationships Redux

While I certainly cannot expect everyone in LIS to be enamored of every one of these types of relationships, I most certainly do expect every LIS "professional" to be concerned with the kinds that most directly impinge on their particular area(s) of expertise.

Relationships are everywhere.  There is no reality without them, at least not a reality processable by humans or any other form of life as we understand it.  Seeing as LIS is concerned with the recorded forms of human knowledge, they are inescapable.

We have been obsessed with "entities," things, for far too long.  Perhaps it is time we pay more attention to what it is that allows us to recognize any entity in the first place.

[any material that should appear in print but not on the slide]

Bibliography 1

Bean, Carol A. and Rebecca Green, Eds. Relationships in the Organization of Knowledge. Information Science and Knowledge Management, v. 2. Dordrecht: Kluwer Academic Publishers, 2001.

IFLA Study Group on the Functional Requirements for Bibliographic Records. (1998). Functional Requirements for Bibliographic Records - Final Report [pdf version]

Green, Rebecca. (1995). "Topical relevance relationships: Why topic matching fails." Journal of the American Society for Information Science 46: 646-653.

Green, Rebecca. (2000). "Locating sources in humanities scholarship: The efficacy of following bibliographic references." Library Quarterly 70: 201-229.

[Bibliography page 1]

Bibliography 2

Green, Rebecca and Carol A. Bean. (1995). "Topical relevance relationships: An exploratory study and preliminary typology." Journal of the American Society for Information Science 46: 654-662.

Green, Rebecca, Carol A. Bean and Sung Hyon Myaeng, Eds. The Semantics of Relationships: An Interdisciplinary Perspective. Information Science and Knowledge Management, v. 3. Dordrecht: Kluwer Academic Publishers, 2002.

Khoo, Christopher and Jin-Cheon Na. (2006). "Semantic relations in information science." Annual Review of Information Science and Technology 40. Medford, NJ: Information Today, p. 157-228.

[Bibliography page 2]

Bibliography 3

Perreault, Jean. (1965) "Categories and relators: a new schema." Reprinted in: Knowledge Organization 21 (4), 1994: 189-197.

Smiraglia, Richard. P. (1992). Authority control and the extent of derivative bibliographic relationships. Ph.D. dissertation, University of Chicago.

Tillett, Barbara B. (1987). Bibliographic relationships: Toward a conceptual structure of bibliographic information used in cataloging. Ph.D. dissertation, University of California, Los Angeles.

Vellucci, Sherry L. (1997). Bibliographic relationships in music catalogs. Lanham, MD: Scarecrow Press.

[Bibliography page 3]

Bibliography 4

Vellucci, Sherry L. (1998) "Bibliographic relationships." The Principles and future of AACR: Proceedings of the International Conference on the Principles and Future Development of AACR, Toronto, Ontario, Canada, October 23-25, 1997. Chicago: American Library Association.

[Bibliography page 4]

Technology Credits

  • S5: A Simple Standards-Based Slide Show System based entirely on XHTML, CSS, and JavaScript.
  • It is the creation of Eric Meyer.
  • It belongs to the Public Domain.
  • Is small, lightweight and will run on any current standards-based system.
  • Further info and download available at http://www.meyerweb.com/eric/tools/s5/