[table of contents for nonframes version]

ANALYZE

Description

The ANALYZE command is used to extract terms, excluding stopwords, from an L-number and to provide statistical analysis of the terms.

Diagram

This diagram illustrates extraction of author names from an answer set using ANALYZE.

---

Command Line	To use the ANALYZE command, follow this sample format:
	=>	ANALYZE	L-number	extraction field	answer or term numbers
		1	2	3	4

Example: => ANALYZE L3 AU 1-

Command Guidelines

Use the guidelines below to enter command line information:

Field	Guidelines
1	Type: ANALYZE
2	Type: the L-number of an answer set An L-number created with ANALYZE Default: The last L-number created
3	What you type in Field 3 depends on what you type in Field 2.
	IF, in Field 2 you typed	THEN, in Field 3
	An answer set L-number	Type up to 5 extraction field codes. Default: The default extraction field of the file
	An L-number created with ANALYZE	Do not type anything
4	What you type in Field 4 depends on what you type in Field 2.
	IF, in Field 2 you typed	THEN, in Field 4
	An answer set L-number	Type: Example: A single answer number 3 Multiple answer numbers 1 5 7 A range of answer numbers 1-10 A combination of these 1-10 17 All the answers 1- Default: All answers
	An L-number created with ANALYZE	Type: Example: A single term number 1 Multiple term numbers 2,4 A range of term numbers 1-10 A combination of these 1-10, 15 All the answers 1- A range criteria OGT 100 Range Criteria Definitions: TOP n = the first n terms based on current ordering OGT n = the terms with occurrence counts greater than n DGT n = the terms with document counts greater than n PGT n = the terms with percent of document counts greater than n Default: All terms

Note: An L-number can only be ANALYZEd in the same file in which it was created.

Restricting Extracted Terms

You may use ANALYZE to extract terms only if they meet certain criteria.

IF, you want to extract	THEN, to the command line add	Example
Terms that contain a 1-20 character string	WITH followed by the character string in quotes	=> ANALYZE L5 AU 1-10 WITH "BROWN"
Terms that do not contain a 1-20 character string	NOT followed by the character string in quotes	=> ANALYZE L9 TI 1- NOT "METHANE"
The first n characters of each term	LEN n	=> ANALYZE L6 1- LEN 10
Rules: When extracting from an ANALYZE L-number, WITH and NOT apply to both the term and the appended field code. WITH and NOT may not be used in the same ANALYZE command. LEN may be used with WITH or with NOT in one ANALYZE command.

Results

Results obtained with the ANALYZE command are placed in a new L-number. Use the DISPLAY command with the ANALYZE L-number to view: Selected terms The number of times each term occurs The number of documents that contain each term The percentage of documents that contain each term Specialized displays of ANALYZE results are available. The ANALYZE L-number may be used: As a search term in subsequent SEARCH commands In subsequent ANALYZE or TABULATE commands

Example

The following example shows use of ANALYZE to extract and perform statistical analysis on authors from answer set L1.

=> FILE GEOREF => S VENUS AND (VOLCAN? OR OUTGAS? OR DEGAS?) L1 917 VENUS AND (VOLCAN? OR OUTGAS? OR DEGAS?) => ANALYZE L1 AU 1- L2 ANALYZE L1 1- AU : 815 TERMS => D L2 L2 ANALYZE L1 1- AU : 815 TERMS TERM # # OCC # DOC % DOC AU ------ ------- ------ ------ --------------- 1 91 91 9.92 HEAD, JAMES W. 2 57 57 6.22 HEAD, J. W. 3 34 34 3.71 SOLOMON, SEAN C. 4 29 29 3.16 CRUMPLER, L. S. 5 25 25 2.73 BASILEVSKY, A. T. 6 24 24 2.62 HEAD, JAMES W., III 7 24 24 2.62 SAUNDERS, R. STEPHEN 8 23 23 2.51 STOFAN, ELLEN R. 9 21 21 2.29 GUEST, J. E. 10 21 21 2.29 WILSON, LIONEL

Limits

50,000 terms may be extracted with ANALYZE. There is a limit of 1000 terms per answer.

STN Assistance

If you enter just the ANALYZE command, STN prompts you for: An answer set or ANALYZE L-number A term range An display (extraction) field code If you are unsure how to answer an STN prompt, enter a question mark (?) and STN provides more details. For more details on the ANALYZE command, type HELP ANALYZE.