Group:
natural language
Topic:
data record
Topic:
dictionary for natural language
Topic:
graphs
Topic:
hypertext as external memory
Topic:
intertextuality and hypertext
Topic:
knowledge as interrelated facts
Topic:
knowledge representation
Topic:
meaning by language as a whole
Topic:
meaning of words
Topic:
program web
Topic:
thought is computational
Topic:
Thesa data model
Topic:
thesaurus and information retrieval
Topic:
using a computer as a communication/information medium
Topic:
words in natural languages
| |
Summary
A natural way to represent meaning is through semantic networks of concepts and their interrelationships, as in a thesaurus or hypertext. The emphasis is on association and relationship. The main alternatives are indexing and hierarchical structures as found in table of contents, index, and dictionary.
A concept in a semantic network is a neighborhood of nodes. Each node acts as the root of a collection of information. (cbb 4/07)
Subtopic: semantic net as meaning 
| Quote: meaning (noemata) consists of layers of predicate-senses with hierarchical relations and cross-links; organizes one's experience [»hallH_1992]
| | Quote: must be a notation for representing the meaning of sentences independent of word sequence [»woodWA_1975]
| | Quote: abstraction is a visible reference to something in the mind; a node in an information web [»cbb_1973, OK]
| | Quote: conceptual graphs are a universal, language-independent deep structure; a concept is a node of the graph, a unit of knowledge [»sowaJF_1984]
| | Quote: concept structure represented by a computerized symbol structure displayed as a two-dim slice [»engeDC_1963]
| | Quote: a frequently used name is associated with a rich collection of intentional contents in the Network
|
Subtopic: concept as reachable nodes
| Quote: a concept in a semantic net is a hierarchically ordered, extensive body of information; from general facts to obscure [»quilMR_1967]
| | Quote: a concept in a semantic net is all nodes reachable from its type node and their interrelationships [»quilMR_1967]
| | Quote: a type node in a semantic net is linked to a configuration of other nodes that represent the meaning of its name word [»quilMR_1967]
| | Quote: a definition of a semantic net concept is a plane of token nodes interlinked into a configuration; each token is linked to another plane [»quilMR_1967]
| | Quote: example of intersecting 'cry' and 'comfort' in a semantic net [»quilMR_1967]
|
Subtopic: semantic net as model
| Quote: any collection of knowledge about something can be modeled to any degree of detail by an appropriate plex; nothing simpler; foundation of the Computer-Aided Design System [»rossDT_1963]
| | Quote: a plex is an interconnect set of n-component elements; represents arbitrarily complex information from any source; e.g., a line has a type, a name, and its two end points
| | Quote: plex structures become so elaborate and interwoven that they are impossible to understand; need to automatically translate from language to plex structures [»rossDT_1963]
| | Quote: use reference types to link records into structural networks of any desired complexity [»wirtN6_1966]
|
Subtopic: semantic nets as interconnected nodes
| Quote: MUCH forms a semantic net with nodes and links given semantically meaningful labels and plausible connectivities [»radaR3_1993]
| | Quote: a semantic net is a complex network of attribute-value nodes with labeled associations between them; including subclass and superclass links [»quilMR_1967]
| | Quote: a frame is a network of nodes and relations; made of slots that usually reference smaller subframes
| | Quote: semantic net model of memory consists of nodes interconnected by associative links; can think of nodes with English name words [»quilMR_1967]
| | Quote: Connection Machine motivated by semantic networks; e.g., processor for 'apple' connected to processor for 'red' [»hillWD_1985]
|
Subtopic: no roots or primitives
| Quote: in semantic nets, no predefined class hierarchy; every word is the root of its own hierarchy [»quilMR_1967]
| | Quote: in semantic nets, there are no primitive word concepts; everything defined in terms of other things in the memory [»quilMR_1967]
| | Quote: in a semantic net, a token node will have many links to it; some circular [»quilMR_1967]
|
Subtopic: links as relationship
| Quote: use references to represent functional relationships between objects; e.g., the 'father' field of A points to A's father [»wirtN6_1966]
|
Subtopic: links as semantic regularity
| Quote: the connotation of an element is the set of similarities noted by hearers of the language about its semantic regularities in the language [»ziffP_1960]
|
Subtopic: hypertext as semantic web
| Quote: hypertext is a semantic network that mixes informal text with mechanized operations [»conkJ9_1987]
| | Quote: hypertext models the deep structure of human idea processing; allows three-dimensional navigation through information [»carlPA_1988]
| | Quote: hypertext nodes can correspond to objects in the world [»conkJ9_1987]
| | Quote: hypertext captures the deep structure of information; best for complex information or information with imprecise internal relationships [»carlPA_1988]
|
Subtopic: thesaurus as semantic net
| Quote: a thesaurus represents the relations between word senses rather than words; be careful when expanding query words with a thesaurus [»krovR4_1992]
| | Quote: WordNet captures the meaning of a word form by a set of synonyms (synset), a gloss, and relationships to other synsets [»millGA7_1990]
| | Quote: WordNet is like an on-line thesaurus (synsets of synonyms) with short glosses and related words; antonym, hyponym, cause, etc. [»beckR7_1990]
|
Subtopic: type vs. token
| Quote: a token node in a semantic net has an associative link to that concept's type node [»quilMR_1967]
|
Subtopic: reading and writing
| Quote: readers decode words into hierarchy into network [»smitJB6_1987]
| | Quote: WE sees reading and writing as transforming linear, hierarchy, and network representations [»conkJ9_1987]
| | Quote: writers begin by bottom-up thinking by building a semantic network [»smitJB6_1987]
|
Subtopic: hierarchy vs. network
| Quote: hierarchical filing doesn't work since the mind uses associations [»bushV7_1945]
|
Subtopic: meaning is not a network
| Quote: since the world is a continuum, a network of concepts is at best a workable approximation [»sowaJF_1984]
|
Subtopic: problems with IS-A links
| Quote: although IS-A links are widely used for classification, they have different meanings in different knowledge-representation systems [»bracRJ10_1983]
| | Quote: using IS-A links for default properties is insufficient for complex concepts and interacts badly with cancellation [»bracRJ10_1983]
| | Quote: the meaning of an IS-A link depends on understanding its nodes [»bracRJ10_1983]
| Quote: IS-A links should be independent of inheritance; inheritance is a implementation issue
[»bracRJ10_1983]
|
Related Topics
Group: natural language (16 topics, 539 quotes)
Topic: data record (57 items)
Topic: dictionary for natural language (41 items)
Topic: graphs (18 items)
Topic: hypertext as external memory (24 items)
Topic: intertextuality and hypertext (11 items)
Topic: knowledge as interrelated facts (23 items)
Topic: knowledge representation (41 items)
Topic: meaning by language as a whole (26 items)
Topic: meaning of words (21 items)
Topic: program web (8 items)
Topic: thought is computational (60 items)
Topic: Thesa data model (58 items)
Topic: thesaurus and information retrieval (29 items)
Topic: using a computer as a communication/information medium (32 items)
Topic: words in natural languages
(40 items)
|
