Acronym Extraction System and Method of Identifying Acronyms and Extracting Corresponding Expansions from Text

• US 20080033714A1
• Filed: 06/14/2007
• Published: 02/07/2008
• Est. Priority Date: 08/07/2002
• Status: Abandoned Application

• Current Assignee: ITT Manufacturing Enterprises, Inc. (ITT, Inc.)
• Original Assignee: ITT Manufacturing Enterprises, Inc. (ITT, Inc.)
• Inventors: Gupta, Kalyan
• Application Number:
  US11/762,816
• Publication Number:
  US 20080033714A1
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates

Abstract

An acronym expansion system of the present invention receives electronic documents and extracts acronyms and their corresponding expansions. A part-of-speech tagger decomposes text into string tokens or words and tags them with their part-of-speech, while an acronym identifier determines whether a word is a potential acronym based on various conditions. An expansion identifier retrieves lists of words preceding and following a potential acronym to search for the expansion. The resulting word lists are examined sequentially to identify and retrieve an expansion for the potential acronym. An expansion extractor receives the potential acronym and a processed word list to retrieve the expansion of the potential acronym from that list. The extractor may utilize information from prior search iterations, and verifies an extracted expansion against a set of rules to remove spurious expansions.

Claims

1-95. -95. (canceled)

96. A system for identifying abbreviated terms within text each representing a corresponding phrase of at least one term and extracting expansions of said abbreviated terms from said text in the form of said corresponding phrases comprising:

a computer system to receive said text and identify abbreviated terms and corresponding expansions therein, said computer system including;

an identification module to examine said text to identify at least one abbreviated term residing therein;

an expansion retrieval module to retrieve at least one portion of said text for an identified abbreviated term, wherein each retrieved text portion is located within said text proximate said identified abbreviated term; and

an expansion extraction module to produce a plurality of sets of terms from a corresponding retrieved text portion with each set including a member term that includes an initial portion containing an initial portion of said identified abbreviated term, to compare said identified abbreviated term with said produced sets to extract an expansion for said abbreviated term from one of said sets, and to verify said extracted expansion to produce a valid expansion for said identified abbreviated term.

97. The system of claim 96, wherein said at least one abbreviated term includes an acronym.

98. The system of claim 96 further including:

a classification module to determine a grammatical classification of each term within said text to enable identification of an abbreviated term and a corresponding valid expansion.

99. The system of claim 98, wherein said identification module includes:

a text attribute module to examine a term within said text and determine compliance of text term attributes with particular conditions in order to identify that text term as an abbreviated term, wherein said conditions relate to at least one of;

said grammatical classification of a text term;

a length of a text term being within a particular range;

the presence of a predetermined delimiter within a text term;

an amount of capitalization within a text term;

an amount of numerals within a text term; and

a text term being a predetermined term excluded from consideration as an abbreviated term.

100. The system of claim 96 further including:

a decomposition module to selectively decompose terms within at least one retrieved text portion for said identified abbreviated term into individual sub-terms to facilitate identification of an expansion within that retrieved text portion.

101. The system of claim 96, wherein said expansion extraction module includes:

a set module to identify terms within each retrieved text portion that include an initial portion containing a first character of said identified abbreviated term and to produce a corresponding set for each identified term including said identified term and following terms within said retrieved text portion.

102. The system of claim 101, wherein said expansion extraction module further includes:

a search module to compare portions of said identified abbreviated term to terms within a search window to identify a corresponding expansion term for each abbreviated term portion, wherein said search window includes a predetermined number of terms from a corresponding set and is modified in response to identification of an expansion term to include terms within that set following said identified expansion term.

103. The system of claim 102, wherein said search module includes:

an expansion term module to compare an abbreviated term portion to said terms within said search window and identify said corresponding expansion term for that abbreviated term portion in response to a search window term including that abbreviated term portion as a starting portion or within an initial portion of that search window term.

104. The system of claim 102, wherein said search module includes:

a backtrack module to identify a corresponding expansion term for a current abbreviated term portion utilizing at least one previous abbreviated term portion and at least one term from a previous search window in response to a failure to identify an expansion term for said current abbreviated term portion.

105. The system of claim 104, wherein said backtrack module includes:

a first level module to compare a combination of the current and previous portions of said identified abbreviated term to terms within a previous search window to identify an expansion term for said current abbreviated term portion.

106. The system of claim 105, wherein said backtrack module further includes:

a second level module to compare a combination of the current and a plurality of previous portions of said identified abbreviated term to a combination of terms including terms within a current search window and a plurality of previous identified expansion terms to identify an expansion term for said current abbreviated term portion in response to a failure by said first level module to identify that expansion term.

107. The system of claim 96, wherein said expansion extraction module includes:

a verification module to verify said extracted expansion against a set of verification rules that indicate a valid expansion in response to attributes of said extracted expansion satisfying particular conditions.

108. The system of claim 107, wherein said conditions relate to at least one of:

said extracted expansion including said identified abbreviated term, a predetermined delimiter or a predetermined expansion invalidating term;

numerals within said identified abbreviated term corresponding to numerals within said extracted expansion;

a first term of said extracted expansion including a predetermined expansion invalidating term;

a grammatical classification of a last term of said extracted expansion;

the presence of particular terms within said extracted expansion being skipped during identification of said extracted expansion; and

the presence of at least two terms within said extracted expansion containing an abbreviated term portion within an intermediate portion of those terms.

109. A method of identifying abbreviated terms within text each representing a corresponding phrase of at least one term and extracting expansions of said abbreviated terms from said text in the form of said corresponding phrases comprising:

(a) examining said text to identify at least one abbreviated term residing therein;

(b) retrieving at least one portion of said text for an identified abbreviated term, wherein each retrieved text portion is located within said text proximate said identified abbreviated term; and

(c) producing a valid expansion for said identified abbreviated term by producing a plurality of sets of terms from a corresponding retrieved text portion with each set including a member term that includes an initial portion containing an initial portion of said identified abbreviated term, comparing said identified abbreviated term with said produced sets to extract an expansion for said abbreviated term from one of said sets, and verifying said extracted expansion.

110. The method of claim 109, wherein said at least one abbreviated term includes an acronym.

111. The method of claim 109, wherein step (a) further includes:

(a.1) determining a grammatical classification of each term within said text to enable identification of an abbreviated term and a corresponding valid expansion.

112. The method of claim 111, wherein step (a.1) further includes:

(a.1.1) examining a term within said text and determining compliance of text term attributes with particular conditions in order to identify that text term as an abbreviated term, wherein said conditions relate to at least one of;

said grammatical classification of a text term;

a length of a text term being within a particular range;

the presence of a predetermined delimiter within a text term;

an amount of capitalization within a text term;

an amount of numerals within a text term; and

a text term being a predetermined term excluded from consideration as an abbreviated term.

113. The method of claim 109, wherein step (b) further includes:

(b.1) selectively decomposing terms within at least one retrieved text portion for said identified abbreviated term into individual sub-terms to facilitate identification of an expansion within that retrieved text portion.

114. The method of claim 109, wherein step (c) further includes:

(c.1) identifying terms within each retrieved text portion that include an initial portion containing a first character of said identified abbreviated term and producing a corresponding set for each identified term including said identified term and following terms within said retrieved text portion.

115. The method of claim 114, wherein step (c) further includes:

(c.2) comparing portions of said identified abbreviated term to terms within a search window to identify a corresponding expansion term for each abbreviated term portion, wherein said search window includes a predetermined number of terms from a corresponding set and is modified in response to identification of an expansion term to include terms within that set following said identified expansion term.

116. The method of claim 115, wherein step (c.2) further includes:

(c.2.1) comparing an abbreviated term portion to said terms within said search window and identifying said corresponding expansion term for that abbreviated term portion in response to a search window term including that abbreviated term portion as a starting portion or within an initial portion of that search window term.

117. The method of claim 115, wherein step (c.2) further includes:

(c.2.1) identifying a corresponding expansion term for a current abbreviated term portion utilizing at least one previous abbreviated term portion and at least one term from a previous search window in response to a failure to identify an expansion term for said current abbreviated term portion.

118. The method of claim 117, wherein step (c.2.1) further includes:

(c.2.1.1) comparing a combination of the current and previous portions of said identified abbreviated term to terms within a previous search window to identify an expansion term for said current abbreviated term portion.

119. The method of claim 118, wherein step (c.2.1) further includes:

(c.2.1.2) comparing a combination of the current and a plurality of previous portions of said identified abbreviated term to a combination of terms including terms within a current search window and a plurality of previous identified expansion terms to identify an expansion term for said current abbreviated term portion in response to a failure to identify that expansion term within step (c.2.1.1).

120. The method of claim 109, wherein step (c) further includes:

(c.1) verifying said extracted expansion against a set of verification rules that indicate a valid expansion in response to attributes of said extracted expansion satisfying particular conditions.

121. The method of claim 120, wherein said conditions relate to at least one of:

said extracted expansion including said identified abbreviated term, a predetermined delimiter or a predetermined expansion invalidating term;

numerals within said identified abbreviated term corresponding to numerals within said extracted expansion;

a first term of said extracted expansion including a predetermined expansion invalidating term;

a grammatical classification of a last term of said extracted expansion;

the presence of particular terms within said extracted expansion being skipped during identification of said extracted expansion; and

the presence of at least two terms within said extracted expansion containing an abbreviated term portion within an intermediate portion of those terms.

122. A program product apparatus including a computer readable medium with computer program logic recorded thereon for identifying abbreviated terms within text each representing a corresponding phrase of at least one term and extracting expansions of said abbreviated terms from said text in the form of said corresponding phrases, said program product apparatus comprising:

an identification module to examine said text to identify at least one abbreviated term residing therein;

an expansion retrieval module to retrieve at least one portion of said text for an identified abbreviated term, wherein each retrieved text portion is located within said text proximate said identified abbreviated term; and

an expansion extraction module produce a plurality of sets of terms from a corresponding retrieved text portion with each set including a member term that includes an initial portion containing an initial portion of said identified abbreviated term, to compare said identified abbreviated term with said produced sets to extract an expansion for said abbreviated term from one of said sets, and to verify said extracted expansion to produce a valid expansion for said identified abbreviated term.

123. The apparatus of claim 122, wherein said at least one abbreviated term includes an acronym.

124. The apparatus of claim 122 further including:

a classification module to determine a grammatical classification of each term within said text to enable identification of an abbreviated term and a corresponding valid expansion.

125. The apparatus of claim 122, wherein said identification module includes:

a text attribute module to examine a term within said text and determine compliance of text term attributes with particular conditions in order to identify that text term as an abbreviated term, wherein said conditions relate to at least one of;

said grammatical classification of a text term;

a length of a text term being within a particular range;

the presence of a predetermined delimiter within a text term;

an amount of capitalization within a text term;

an amount of numerals within a text term; and

a text term being a predetermined term excluded from consideration as an abbreviated term.

126. The apparatus of claim 122 further including:

a decomposition module to selectively decompose terms within at least one retrieved text portion for said identified abbreviated term into individual sub-terms to facilitate identification of an expansion within that retrieved text portion.

127. The apparatus of claim 122, wherein said expansion extraction module includes:

a set module to identify terms within each retrieved text portion that include an initial portion containing a first character of said identified abbreviated term and to produce a corresponding set for each identified term including said identified term and following terms within said retrieved text portion.

128. The apparatus of claim 127, wherein said expansion extraction module further includes:

a search module to compare portions of said identified abbreviated term to terms within a search window to identify a corresponding expansion term for each abbreviated term portion, wherein said search window includes a predetermined number of terms from a corresponding set and is modified in response to identification of an expansion term to include terms within that set following said identified expansion term.

129. The apparatus of claim 128, wherein said search module includes:

an expansion term module to compare an abbreviated term portion to said terms within said search window and identify said corresponding expansion term for that abbreviated term portion in response to a search window term including that abbreviated term portion as a starting portion or within an initial portion of that search window term.

130. The apparatus of claim 128, wherein said search module includes:

a backtrack module to identify a corresponding expansion term for a current abbreviated term portion utilizing at least one previous abbreviated term portion and at least one term from a previous search window in response to a failure to identify an expansion term for said current abbreviated term portion.

131. The apparatus of claim 130, wherein said backtrack module includes:

a first level module to compare a combination of the current and previous portions of said identified abbreviated term to terms within a previous search window to identify an expansion term for said current abbreviated term portion.

132. The apparatus of claim 131, wherein said backtrack module further includes:

a second level module to compare a combination of the current and a plurality of previous portions of said identified abbreviated term to a combination of terms including terms within a current search window and a plurality of previous identified expansion terms to identify an expansion term for said current abbreviated term portion in response to a failure by said first level module to identify that expansion term.

133. The apparatus of claim 122, wherein said expansion extraction module includes:

a verification module to verify said extracted expansion against a set of verification rules that indicate a valid expansion in response to attributes of said extracted expansion satisfying particular conditions.

134. The apparatus of claim 133, wherein said conditions relate to at least one of:

said extracted expansion including said identified abbreviated term, a predetermined delimiter or a predetermined expansion invalidating term;

numerals within said identified abbreviated term corresponding to numerals within said extracted expansion;

a first term of said extracted expansion including a predetermined expansion invalidating term;

a grammatical classification of a last term of said extracted expansion;

the presence of particular terms within said extracted expansion being skipped during identification of said extracted expansion; and

the presence of at least two terms within said extracted expansion containing an abbreviated term portion within an intermediate portion of those terms.