Introduction

К оглавлению
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 
34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 
51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 
68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 
102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 
136 137 138 139 140 141 142 143 144 145 146 

Texts are a typical source of information about meaning, organizations, and society.

Today, a large and growing number of texts are available in an electronic form that

describes, discusses, or displays information about people, the groups to which they

belong, the activities in which they engage, and the resources at their disposal. This data

and its accessibility motivate the development and investigation of automated techniques

for extracting the underlying social and organizational structure from such texts

in an effective and efficient way.

In this chapter, we present an automated approach to text analysis that can be used to

extract the underlying social and organizational structure contained in texts. This

approach is based on the following insights. First, texts can be represented as networks

of concepts and the connections between them. These concepts refer to ideas, people,

resources, organizations, events, etc. Second, many of the items referred to, such as

people, are core entities in the structure of groups and organizations. Hence, the

extracted networks contain representations of the social structure — the entities and

relations among them that comprise a group, organization, or society. By classifying the

concepts into entity classes used in defining social structures and partitioning the

extracted network into sub-networks, we have effectively used network analysis of texts

to reveal the social structure represented in texts.

Herein, we describe this approach in detail and explain how we operationalized, formalized,

and implemented it into a software called AutoMap that enables analysts to extract

social structure from texts. As part of this work, we have operationalized an ontological

scheme based on the meta-matrix proposed by Carley (2002) for describing social and

organizational structure. This ontology is utilized as part of a hierarchical scheme for

cross categorizing concepts. In this chapter we furthermore demonstrate how analysts

can use AutoMap to automatically extract not just networks of concepts and the relations

among them, but also classify the concepts and relations between them according to this

ontology. This enables the automatic extraction of views of the social structure.

The chapter begins with a brief overview on the model and methods involved. We then

describe how we formalized and implemented the combination of the meta-matrix model

and the network text analysis technique. This is followed by a substantive example that

we provide in order to illustrate this approach for revealing social structure through the

analysis of texts by extracting an image of the social structure of the terrorists groups

in the West Bank. We conclude with a discussion of the potentials and limitations of our

approach.

Note, this chapter should not be viewed as a description of the West Bank terrorist

groups. We have coded for this chapter only a small sample of texts to illustrate the

technique. No conclusions for this group should be drawn from the results reported

herein.

Revealing Social Structure from Texts 83