Indexing subject-locating method

Claim 2

An automated method for locating in a textual manuscript the subjects of a list of subjects, said method comprising the steps of deciding for each word of a predetermined segment of text whether or not it is contained in any word position in at least one subject of said subject list, determining for each subject containing a segment word whether or not all other words of said each subject are contained in a portion of said manuscript of predetermined size, and including at least the same segment, recording a subject, found in said determining step to have all of its words in a segment, with an indication of the location of such subject and segment in said manuscript, and repeating said deciding, determining, and recording steps for at least partially different segments of text.

Claim 3

The subject-locating method in accordance with claim 1 in which said recording step comprises the steps of providing a text processing system paging command for associating a manuscript page number with a selected subject referenced by said command, and storing with each segment both each subject located therein and a copy of said paging command.

Claim 4

The subject-locating method in accordance with claim 1 in which there is included a segment-selecting step comprising reading a character of said manuscript, testing said character to establish whether it is a type of character either indicating a text word or indicating an effective end of an n-sentence segment, providing each text word, established in said testing step, to said deciding step, and interrupting said providing step, in response to an end-of-segment character, and transferring from said deciding step to said determining step.

Claim 5

The subject-locating method in accordance with claim 1 in which there is included a segment-selecting step comprising the steps of reading a token portion of said text manuscript, and testing said token portion to determine which of predetermined plural types it represents, and if said token is a word, executing said deciding step, if said token is an end-of-sentence token, testing whether n sentences have yet been read, and, if not, get a new token, but, if true, go to said determining step, if said token is a heading or a paragraph token, go to said determining step, and if said token is an end-of-manuscript-file token, set an end-of-file flag to stop said repeating step, and go to said determining step.

Claim 6

The subject-locating method in accordance with claim 1 in which each subject of said list has a match-type indicator specifying at least one predetermined degree of specificity required for word match operations used to locate the subject in said manuscript, said deciding step includes the steps of comparing each manuscript word to words of subjects of said subject list, each comparison including successive matching operations of different specificity types, and recording for each match operation that produces an affirmative result both the manuscript word involved and the type of match used to produce that result, and said determining step includes the step of accepting for said subject recording step those subjects for which all subject words recorded in said word recording step have match-types at least as specific as a match-type required by the subject to be accepted for the corresponding subject word.

Claim 7

The subject-locating method in accordance with claim 1 in which said deciding step comprises the steps of constructing a target word list containing only once each word of any subject of said subject list, activating each target word which is included in said segment of text, said activating function comprising the step of storing a pointer to such target word in an active word list, and providing said each target word activated in said activating step as input to said determining step.

Claim 8

The subject-locating method in accordance with claim 6 in which said constructing step includes the step of storing each target word with a pointer to each subject of said subject list including said each target word, and said determining step comprises the steps of selecting a word from said active word list, finding a corresponding word on said target word list and selecting an associated pointer to said subject list, testing the pointed subject for the presence of all words thereof on said active list and marking the tested subject with its test result, repeating the step of testing for any other subjects that are pointed from the same target list word, and repeating the steps of selecting, finding, testing, and repeating testing for each word of a portion of said active list for a sentence of said segment.

Claim 9

The subject-locating method in accordance with claim 6 in which said activating step comprises the steps of testing each word of said segment which is not in said target list for uppercase letters, and if any, converting such word to all lowercase letters, if the lowercase word form just produced in said converting operation is on said target list, storing a pointer to such target word in said active word list, screening each word of said segment to identify words having a predetermined high-indexing-significant suffix, discarding said suffix, testing each word of said segment to determine whether or not its root at least approximately matches a target list word root, and activating any target word so matched.

Claim 10

The subject-locating method in accordance with claim 8 in which said root-activating step comprises the steps of testing said target list for a word having the same first alphabetic character as said word root of said root-match-testing step, and if a same-first-letter word is not found, return to said deciding step, but, if found, testing for at least an approximate root match between said word root and said target list same-first-letter word, activating said target word if a match is found in said step of testing for at least an approximate root match, and whether or not an at least approximate root match occurred, loop back to the step for testing said target list for an additional word having the same first alphabetic character as said word root.

Claim 11

The subject-locating method in accordance with claim 9 in which said testing for an at least approximate root-word-matching step comprises the step of identifying the shorter one of the roots and testing whether or not it is at least a predetermined number of characters in length, and if the shorter root has at least the minimum length, testing for a match between (a) the shorter root and (b) the first N-1 characters of the other of the two roots, where N is the total number of characters of the longer word.

Claim 14

The method in accordance with claim 11 in which words are matched by a literal type of matching process or a less specific type of matching process, at least one subject includes a requirement that a literal match be used to consider that it has been found in said text, and said first determining step includes the steps of using either matching type to determine whether or not said each word is included in any one or more of said subjects, said second determining step further includes the step of determining as to each subject whether or not matching types used in said first determining step and in this second determining step are at least as specific as a match type required by said each subject, and said discontinuing step further includes the step of discontinuing said second determining step for a selected segment as to any said each subject for which matching types in said first and second determining steps are not at least as specific as a match type required by said each subject.