Bibexcel is designed as a tool box for manipulating bibliographic data. The result of your manipulations will be saved in files that can be opened with Excel or any other software reading text-files tabbed into columns. Bibexcel let's you combine information from several fields of a document record, count frequencies, co-occurrences and shared units (bibliographic coupling). Among other things there is also a procedure for finding citation links among the documents within a given set. Above all, the tools can be combined - the result of using them depends far more on your own imagination than the tools themselves.
To understand what Bibexcel can be used for imagine the following situation.
You have decided to analyze the development of memory research. Some basic questions might be:
Who are the most productive authors, universities or countries?
Which journals are used for publishing memory research?
Using Social Sciences Citation Index, SSCI-CDE (compact disk edition) you have downloaded a set of records that contain the word memory in the title field. A total of 246 records are saved in a file called memory.doc. The first record looks like this:
FN- Social Sciences Citation Index (Jan 81 - Dec 85)
GA- AHQ96|
TI- MEMORY ACCESSIBILITY AND TASK INVOLVEMENT AS FACTORS IN
CHOICE|
LA- ENGLISH|
AU- GARDIAL SF; BIEHAL GJ|
CS- UNIV HOUSTON/HOUSTON//TX/77004|
JN- ADVANCES IN CONSUMER RESEARCH, 1985, V12, P414-419|
PY- 1985|
DT- ARTICLE|
NR- 17|
CR- BATRA R, 1983, V10, P309, ADV CONSUM RES
BETTMAN JR, 1979, INFORMATION PROCESSI
BETTMAN JR, 1980, V7, P148, ADV CONSUMER RES
BETTMAN JR, 1980, V7, P234, J CONSUMER RES
BIEHAL G, 1982, V8, P431, J CONSUMER RES
BIEHAL G, 1983, V10, P1, J CONSUM RES
COLLINS AM, 1975, V82, P407, PSYCHOL REV
CRAIK FIM, 1972, V11, P671, J VERB LEARN VERB BE
EAGLE M, 1964, V68, P58, J EXP PSYCHOL
JOHNSON EJ, 1984, V11, P542, J CONSUMER RES
LEAVITT C, 1981, V8, P15, ADV CONSUMER RES
MITCHELL AA, 1981, V8, P25, ADV CONSUMER RES
PETTY RE, 1981, COGNITIVE RESPONSES
PETTY RE, 1983, V10, P135, J CONSUM RES
TULVING E, 1966, V5, P381, J VERB LEARN VERB BE
TULVING E, 1971, V87, P1, J EXP PSYCHOL
TULVING E, 1973, V80, P352, PSYCHOL REV||
The record has a number of fields ending with a spike \"|\with an extra spike at the end of the last field. Each field starts with a tag in the first four columns, \"TI- \" for title etc. A field may have several parts (such as authors) we will call such parts units.
Record and document means the same thing from now on.
To find the most productive authors, Bibexcel will read the AU-field and save that information in a file named memory.out that looks like this (See Preparing the data):
1 BIEHAL GJ
1 GARDIAL SF
2 TULVING E
3 MOSCOVITCH M
4 BACKMAN L
4 HARDY J
4 NILSSON LG
4 WINBLAD B
5 WEINGARTNER H
6 ROBERTS JV
cont......
The first two lines contain the authors of document nr 1, the document number is listed first and then the author names. A tab separates the two columns, which enables you to open the file in Excel.
Also note that the author names within a document are sorted in alphabetic order, to enable more effective analysis later on. But don't worry, if you want to keep the ordering of author-names there are tricks to do it!
And finally, who has published the most? Well, Bibexcel reads memory.out and sorts it by author, then counts the number of occurrences and write the frequencies to a file called memory.cit which look like this (See Frequency distribution):
5 KAUSLER DH
4 KASZNIAK AW
4 ACKERMAN BP
4 WILSON RS
4 GLOVER JA
3 PARKIN AJ
3 JACOBY LL
3 WOLTERS G
3 PUCKETT JM
3 BACON LD
3 TILL RE
3 TULVING E
3 YESAVAGE JA
A various number of other procedures or tools are available both for editing and analyzing your records. In fact there is hardly any limit to what the tools and their combination can achieve.
I do hope that you have got the general idea!
There is much more to learn!
Bibexcel offers a number of tools for extracting and analyzing various fields.
Bibexcel help
Preparing the data/making the outfile
The drive, directory and file list boxes, can be used to select the file that you wish to analyze.
When clicked the selected file appears in the label-box under the file list. This file can be viewed by clicking on the View-button, at least the first 500 rows or so.
You should first select which field to analyze.
Chose one of the options and then press the \"Prep\"-button. When done the program makes an out-file (\"filename.out\"), that for each unit in a field has one row. The row begins with the document number.
\"CR- Cited reference\"
This option makes an out-file with the cited references looking like this:
BATRA R, 1983, V10, P309, ADV CONSUM RES
1 BETTMAN JR, 1979, INFORMATION PROCESSI
1 BETTMAN JR, 1980, V7, P148, ADV CONSUMER RES
1 BETTMAN JR, 1980, V7, P234, J CONSUMER RES
1 BIEHAL G, 1982, V8, P431, J CONSUMER RES
1 BIEHAL G, 1983, V10, P1, J CONSUM RES
1 COLLINS AM, 1975, V82, P407, PSYCHOL REV
1 CRAIK FIM, 1972, V11, P671, J VERB LEARN VERB BE
1 EAGLE M, 1964, V68, P58, J EXP PSYCHOL
1 JOHNSON EJ, 1984, V11, P542, J CONSUMER RES
1 LEAVITT C, 1981, V8, P15, ADV CONSUMER RES
1 MITCHELL AA, 1981, V8, P25, ADV CONSUMER RES
1 PETTY RE, 1981, COGNITIVE RESPONSES
1 PETTY RE, 1983, V10, P135, J CONSUM RES
1 TULVING E, 1966, V5, P381, J VERB LEARN VERB BE
1 TULVING E, 1971, V87, P1, J EXP PSYCHOL
1 TULVING E, 1973, V80, P352, PSYCHOL REV
2 \"Here comes the first reference in document nr 2\".
\"Any ; separated field\"
Some fields have several units, several authors in the author field, or several addresses in the corporate source field. Now, if we want to separate them, we can make one row for each unit.
First you must write the relevant field tag in the box \"Old tag\Then press the \"Prep\"-button.
The filename.out may look like this if you type \"au\" in the 'Old tag'-box:
1 BIEHAL GJ
1 GARDIAL SF
2 TULVING E
3 MOSCOVITCH M
4 BACKMAN L
4 HARDY J
4 NILSSON LG
4 WINBLAD B
5 WEINGARTNER H
6 ROBERTS JV
cont......
636 RAAIJMAKERS JGW
637 ROEDIGER HL
The first document is obviously co-authored.
\"JN- Journal\"
Choosing this will make one row for each document with the journal name:
1 ADVANCES IN CONSUMER RESEARCH
\"Journal, year, vol, page\"
....and this will make a row for the complete journal source data:
1 ADVANCES IN CONSUMER RESEARCH, 1985, V12, P414-419
\"Blank-separated words(e.g. title)\"
...makes an outfile in which each word in title field or any other field is separated, by a blank,
and then sorted:
1 ACCESSIBILITY
1 CHOICE
1 FACTORS
1 INVOLVEMENT
1 MEMORY
1 TASK
2 HOW
2 MANY
2 MEMORY-SYSTEMS
2 THERE
3 IMPLICATIONS
3 INFANCY
3 MEMORY
3 MEMORY
3 NORMAL
3 OLD-AGE
3 PATHOLOGICAL
3 THEORIES
A list of stop words prevents listing of less significant words and characters. If a word is found in this string it will be deleted:
\" a an and are as at be but by for from had have he her his in is it not of on or
that the this to was which with you - 1 2 3 4 5 6 7 8 9 \"
To remove more insignificant words, import the outfile into Excel and use Pivot-table to find word frequencies, or calculate frquencies with Bibexcel. Save the rows containing the words you really want to be included into an unt-file (filename.unt). Keep the frequencies in the left column and the
unit in the second! Then use Analyze/Co-occurence/Select units from file to make the selection.
\"Whole field intact\"
....a row for the whole field. You must set the field tag in \"Old Tag\" first!
This is when AU is wiritten in \"Old Tag\".
1 GARDIAL SF; BIEHAL GJ
Frequency distribution
Below the file list you may choose to calculate frequencies from the outfile, or any other file with the same format, using the various units in the list.
You may also decide if the frequency list should be sorted in descending order by frequency - check Sort descending.
In some cases you may wish to remove duplicates in a field, for example when counting papers by country. If so check Remove duplicates.
The box Min number will ignore frequencies lower than typed in!
When selecting type of unit, you must have a relevant out-file. Some of the options refer to cited reference and thus needs a corresponding out-file, based on the cr-field. Similarly, main organization and country needs an out-file based on the cs- field.
The frequency distribution will be saved in a cit-file (\"filename.cit\").
Supppose you have the memory.out looking like this based on the au-field::
1 BIEHAL GJ
1 GARDIAL SF
2 TULVING E
3 MOSCOVITCH M
4 BACKMAN L
4 HARDY J
4 NILSSON LG
4 WINBLAD B
5 WEINGARTNER H
6 ROBERTS JV
...then memory.cit will look like this (Sort descending is checked, and Whole string is chosen):
5 KAUSLER DH
4 KASZNIAK AW
4 ACKERMAN BP
4 WILSON RS
4 GLOVER JA
3 PARKIN AJ
3 JACOBY LL
3 WOLTERS G
3 PUCKETT JM
3 BACON LD
3 TILL RE
3 TULVING E
3 YESAVAGE JA
Make a new out-file
Bibexcel allows you to make a new out-file using the units defined in the same list that is used for frequency calculations. For example, you may need an out-file with country names instead of whole addresses. Then you just choose \"Country\" from the list, and then check \"Make new outfile\".
If you wish to remove duplicate countries (or any other unit) within a field
check also Remove duplicates\".
The new out-file will have the extension \"filename.oux\".
Suppose you made memory.out using the 'CR-cited reference' option. Then the units of the first document will look like this (one row for each cited reference):
1 BATRA R, 1983, V10, P309, ADV CONSUM RES
1 BETTMAN JR, 1979, INFORMATION PROCESSI
1 BETTMAN JR, 1980, V7, P148, ADV CONSUMER RES
1 BETTMAN JR, 1980, V7, P234, J CONSUMER RES
1 BIEHAL G, 1982, V8, P431, J CONSUMER RES
1 BIEHAL G, 1983, V10, P1, J CONSUM RES
1 COLLINS AM, 1975, V82, P407, PSYCHOL REV
1 CRAIK FIM, 1972, V11, P671, J VERB LEARN VERB BE
1 EAGLE M, 1964, V68, P58, J EXP PSYCHOL
1 JOHNSON EJ, 1984, V11, P542, J CONSUMER RES
1 LEAVITT C, 1981, V8, P15, ADV CONSUMER RES
1 MITCHELL AA, 1981, V8, P25, ADV CONSUMER RES
1 PETTY RE, 1981, COGNITIVE RESPONSES
1 PETTY RE, 1983, V10, P135, J CONSUM RES
1 TULVING E, 1966, V5, P381, J VERB LEARN VERB BE
1 TULVING E, 1971, V87, P1, J EXP PSYCHOL
1 TULVING E, 1973, V80, P352, PSYCHOL REV
If you select memory.out, chose the option 'cited author' and check 'make new outfile' and 'remove duplicates', then the new file memory.oux will look like this(note that duplicate authors Bettman, Biehal, Petty, Tulving have been reduced to one row each):
1 BATRA R
1 BETTMAN JR
1 BIEHAL G
1 COLLINS AM
1 CRAIK FIM
1 EAGLE M
1 JOHNSON EJ
1 LEAVITT C
1 MITCHELL AA
1 PETTY RE
1 TULVING E
Fractionalize
If a field contain several units, for example authors, you may wish to give each unit a fraction. If there are three authors each would get 1/3 as a fraction rather than 1 when counting frequencies. Fractionalization makes the sum of fractions equal to the sum of documents. If you don't fractionalize muli-authored documents you cannot say \"author x has produced 10 out of 100 papers, or 10 percent of the papers\authors in 10 percent of the papers\" or if the base is the sum of authorships \"author x has 10 of, let磗 say 300 authorships\".
If you take the memory.oux above and with 'make new outfile' and 'fractionalize' checked the memory.oux will look like this, in which the sum of fractions will be 1 (in fact 1.001):
1 BATRA R 0.091
1 BETTMAN JR 0.091
1 BIEHAL G 0.091
1 COLLINS AM 0.091
1 CRAIK FIM 0.091
1 EAGLE M 0.091
1 JOHNSON EJ 0.091
1 LEAVITT C 0.091
1 MITCHELL AA 0.091
1 PETTY RE 0.091
1 TULVING E 0.091
or if you would not check make new outfile the memory.cit file would have the following fraction sums sorted in descending order via Excel:
14.442 CRAIK FIM
3.718 TULVING E
1.501 ATKINSON RC
1.559 BADDELEY AD
1.577 HASHER L
2.078 BOWER GH
1.686 JACOBY LL
1.727 KINTSCH W
1.889 PERLMUTTER M
1.926 EYSENCK MW
1.523 BRANSFORD JD
1.403 PAIVIO A
0.558 ROGERS TB
1.025 CERMAK LS
0.554 NEISSER U
0.987 SMITH AD
..compared to non fractionalized frequencies:
243 CRAIK FIM
78 TULVING E
50 BOWER GH
48 BADDELEY AD
47 JACOBY LL
41 KINTSCH W
37 EYSENCK MW
36 UNDERWOOD BJ
35 ANDERSON JR
35 BRANSFORD JD
34 PERLMUTTER M
34 ATKINSON RC
33 HASHER L
30 PAIVIO A
Bibexcel help
Units per record
If you wish to know how many units there are in a given field in the whole doc-file you may use this procedure. If you want to know the degree of co-authorshipsthis procedure takes the out-file (or corresponding file), and counts the number of co-authored papers and also gives information how many papers that has one author, and so on. Results in \"filename.mul\".
Example
If you select memory.doc and then make an out-file choosing \"Any ; separated field\" and type \"AU\" in the \"Old Tag\"-box and press \"Prep\authors is produced. Then select memory.out and go to \"Analyze/Units per record\". memory.mul wil look like this:
N of records with n units
85 1
95 2
42 3
14 4
8 5
2 6
246 docs and 509 units counted
There are 85 records with one author, 95 with two authors and so on.
The memory.mut file will look like this: one row for each document:
1 2
2 1
3 1
4 4
5 1
6 1
Document number 1 has 2 authors, nr 2 and 3 have one etc.
Lotka-like distributions
Although this routine was not initially intended for it, you may create a distribution that tells you how many authors have written one paper, two papers etc. Similarly. taking the memory.cit file containing a list of most cited authors:
243 CRAIK FIM
78 TULVING E
50 BOWER GH
48 BADDELEY AD
47 JACOBY LL
41 KINTSCH W
37 EYSENCK MW
36 UNDERWOOD BJ
35 ANDERSON JR
35 BRANSFORD JD
34 PERLMUTTER M
34 ATKINSON RC
33 HASHER L
30 PAIVIO A
...the memory.mut will look like this, one author being cited in 243 articles and so on to the last row which tells us that 1907 authors have been cited by only 1 article:
243 1
78 1
50 1
48 1
47 1
41 1
37 1
36 1
35 2
34 2
33 1
30 1
29 1
28 1
26 3
23 3
22 4
21 3
20 4
18 4
17 6
16 4
15 4
14 5
13 8
12 10
11 9
10 10
9 23
8 24
7 41
6 37
5 61
4 78
3 147
2 363
1 1901
Using the inverted square-law of productivity invented by Lotka, the number of authors receiving x citations should be proportionally equal to 1/x2. Applying this to our data the estimated distribution comes quite close to the observed:
x Obs. Est.
1 1901 1901
2 363 475
3 147 211
4 78 119
5 61 76
6 37 53
7 41 39
8 24 30
9 23 23
10 10 19
11 9 16
12 10 13
13 8 11
14 5 10
15 4 8
16 4 7
17 6 7
18 4 6
20 4 21 3 22 4 23 3 26 3 28 1 29 1 30 1 33 1 34 2 35 2 36 1 37 1
5
4
4
4
3
2
2
2
2
2
2
1
1
41 1 1
47 1 1
48 1 1
50 1 1
78 1 0
243 1 0
Bibexcel help
co-occurence analysis.
How often do authors collaborate, which are the most co-cited documents, are some journals more co-cited than others?
Such questions call for a co-occurence analysis.
The starting point will be an out-file (\"filename.out\") or any corresponding
file with the same structure as the out-file. Since the number of pairs that can
be formed is exponentially related to the number of documents
and units within them, you should first decide which units to work with.
This can be done in two ways:
Select units via listbox
Put the cit-file (frequencies) in \"The List\" and select the most frequent units
and then you make the pairs by clicking on: Make pairs via listbox.
This example shows the memory.coc file that will contain the pairs of most cited authors:
77 CRAIK FIM TULVING E
50 BOWER GH CRAIK FIM
48 BADDELEY AD CRAIK FIM
47 CRAIK FIM JACOBY LL
41 CRAIK FIM KINTSCH W
37 CRAIK FIM EYSENCK MW
35 ANDERSON JR CRAIK FIM
35 BRANSFORD JD CRAIK FIM
35 CRAIK FIM UNDERWOOD BJ
34 ATKINSON RC CRAIK FIM
34 CRAIK FIM PERLMUTTER M
33 CRAIK FIM HASHER L
29 CERMAK LS CRAIK FIM
29 CRAIK FIM PAIVIO A
28 CRAIK FIM POSTMAN L
Select units via file
You may also prepare a file containing the units. That file has to have the extension:filename.unt
Then you must select the out-file. From that file all units in filename.unt will be selected. Consequently, the new outfile \"filename.uot\" will be shorter.
Then you make the pairs by selecting:
Make pairs via file
This procedure needs a file of the out-type (uot-file is also nice with fewer units).
...with frequencies
means that the number of co-occurences are calculated
...don't count frequencies
means that all pairs with the document numbers are listed. Good to have that option since we may wish to add labels to the pairs.For example, the publication year in which a co-author pair was formed.
The co-occurence result will be saved in a file: filename.coc.
Make a matrix.
You may need to put co-occurrences in matrix form, for example when making a MDS based map.
1. First, view the cit-file to select the units to be included in the matrix
2. Then select the pair-file (freq+tab+pairleft+tab+pairright). It can be a coc-file for co-occurence, but any file with the same format and containing the
relevant units can be used.
3. Then select Make a matrix from the analyze menu. You will be asked if you want to make a lower left matrix or a squared matrix. The matrix will be saved in filename.ma2.
Example on how to make a co-occurence matrix of cited authors
This is an example from the memrory.doc file.
1. Select memory.doc and then select \"CR- cited reference\" and press Prep-button. This will produce memory.out, in which the first document has the following units:
1 BATRA R, 1983, V10, P309, ADV CONSUM RES
1 BETTMAN JR, 1979, INFORMATION PROCESSI
1 BETTMAN JR, 1980, V7, P148, ADV CONSUMER RES
1 BETTMAN JR, 1980, V7, P234, J CONSUMER RES
1 BIEHAL G, 1982, V8, P431, J CONSUMER RES
1 BIEHAL G, 1983, V10, P1, J CONSUM RES
1 COLLINS AM, 1975, V82, P407, PSYCHOL REV
1 CRAIK FIM, 1972, V11, P671, J VERB LEARN VERB BE
1 EAGLE M, 1964, V68, P58, J EXP PSYCHOL
1 JOHNSON EJ, 1984, V11, P542, J CONSUMER RES
1 LEAVITT C, 1981, V8, P15, ADV CONSUMER RES
1 MITCHELL AA, 1981, V8, P25, ADV CONSUMER RES
1 PETTY RE, 1981, COGNITIVE RESPONSES
1 PETTY RE, 1983, V10, P135, J CONSUM RES
1 TULVING E, 1966, V5, P381, J VERB LEARN VERB BE
1 TULVING E, 1971, V87, P1, J EXP PSYCHOL
1 TULVING E, 1973, V80, P352, PSYCHOL REV
2. Select memory.out, select \"Cited author\check \"make new out-file\" and \"remove duplicates\
Start\" and then memory.oux will be made, looking like this with cited
first-author and duplicates removed:
1 BATRA R
1 BETTMAN JR
1 BIEHAL G
1 COLLINS AM
1 CRAIK FIM
1 EAGLE M
1 JOHNSON EJ
1 LEAVITT C
1 MITCHELL AA
1 PETTY RE
1 TULVING E
3. Select memory.oux, select \"Whole string\" and \"Sort descending\press \"Start\" and then memory.cit will contain the frequencies of most cited
first-authors:
243 CRAIK FIM
78 TULVING E
50 BOWER GH
48 BADDELEY AD
47 JACOBY LL
41 KINTSCH W
37 EYSENCK MW
36 UNDERWOOD BJ
35 ANDERSON JR
35 BRANSFORD JD
34 PERLMUTTER M
34 ATKINSON RC
33 HASHER L
30 PAIVIO A
4. View memory.cit and mark the units you wish to calculate co-occurence frequencies for. When marked go to \"Analyze/Co-occurence/Select units via listbox\". When done all other rows will be deleted from the listbox. Then go to the menu: \"Analyze/Co-occurrence/Make pairs via list box\" which will produce memory.coc containing the pairs:
77 CRAIK FIM TULVING E
50 BOWER GH CRAIK FIM
48 BADDELEY AD CRAIK FIM
47 CRAIK FIM JACOBY LL
41 CRAIK FIM KINTSCH W
37 CRAIK FIM EYSENCK MW
35 BRANSFORD JD CRAIK FIM
35 ANDERSON JR CRAIK FIM
5. Go back to view memory.cit and select the units you want to be included in the matrix. Again, go to \"
Analyze/Co-occurence/Select units via listbox\" and only units selected will show in the list. These were my choices:
243 CRAIK FIM
78 TULVING E
50 BOWER GH
48 BADDELEY AD
47 JACOBY LL
6. Select the file memory.coc and then go to \"Analyze/Make matrix\". If you answer positively to make a lower left matrix memory.ma2 will contain this:
BADDELEY AD BOWER GH CRAIK FIM JACOBY LL TULVING E
13
48 50
21 11 47
27 18 77 27
and if a square matrix was chosen this is how memory.ma2 will look:
BADDELEY AD BOWER GH CRAIK FIM JACOBY LL TULVING E
0 13 48 21 27
13 0 50 11 18
48 50 0 47 77
21 11 47 0 27
27 18 77 27 0
For SYSTAT/SPSS users only
This is for Systat users who wish to make an MDS-based map of the matrix. You must select a matrix ma2-file! Then this procedure will make a cmd-file that can be submitted within Systat/Data. The file mapk.dat will hold the MDS-coordinates!
First a new file with the matrix values will be made in c:\\bibexcel and called map.ma2.
The cmd-file will read the map.ma2 file as a similarity matrix with absent diagonal.
The file map.sys will be used by the MDS-module which generates a two-dimnesional scaling of the matrix and the coordinates are finally saved in mapk.dat.
In SPSS you may also load the map.sys file and do the SPSS/MDS from there. An alternative is to load the original filename.ma2 into Excel, copy the matrix values and paste them into the SPPS spread-sheet.
The cmd-file
Save map (This is the Systat sys-file)
Input F1,F2,F3,F4,F5,F6,F7,F8,F9,F10,F11,F12,F13,F14,F15,F16,F17,F18
type=similarity
diagonal = absent
get map.ma2
run
mds
use map
save mapk/config
Scale
Data
use mapk
put mapk
run
Then mapk.dat kan be imported to Excel and in combination with the labels in filename.ma2 a plot can be made to get a map of the co-occurence structure.
Cluster pairs
A quite nice home-made routine for clustering of pairs, sometimes called Persson's Party Clustering.
Imagine you have the following list of pairs:
10 A B
9 D F
8 B C
7 A C
6 F G
5 H I
4 A H
Note that the pairs are sorted by a co-occurence frequency in column 1,
and that the pair-list also must have tabs separating the two units of a pair.
The clustering routine will have the following sequence of events:
(the pairs are invited to a party):
A-B comes first, have to wait in hall
D-F comes next, have to wait in hall.
B-C comes next, forms a cluster-table with A-B in the room
A-C comes next, will be deleted since A-C already in room
F-G comes next, will not find a friend in room
Goes to the hall and finds D-F and
Then D-F-G will form a cluster in the room
H-I comes next, have to wait in hall
A-H comes next, H will cluster with A-B-C
then searches the hall and finds H-I
I will cluster with A-B-C-H
Cluster 1 will hold : A-B-C-H-I
Cluster 2 will hold D-F-G
In this particlar example the clusters will not change if you resort the pair list.
(You may try out that yourself by saving the pairlist above in a text-file named filename.coc and then run Analyze/Co-occurence/Cluster pairs.)
The cluster result will be saved in a file with the extension \"filename.per\".
The selected pair-file must have the pairs arranged so that the first part of the pair is in the second column and the second part of the pair in the third column. The first column may contain the frequency of co-occurence, and the file can be sorted according to that column since that order will determine the order in which the pairs are clustered.
Make citation links.
The documents in your doc-file may cite each other. To find that out, use the procedure
Make citation links.
It starts by making a search key from the records, by taking the :
last name of the first author, publication year, volume number,and starting page.
AU- GARDIAL SF; BIEHAL GJ|
CS- UNIV HOUSTON/HOUSTON//TX/77004|
JN- ADVANCES IN CONSUMER RESEARCH, 1985, V12, P414-419|
Example from the doc above, the search key is: GARDIAL , 1985, V12, P414.
This search key will free us from spelling variants of initials and cited journal abbreviations, enabling a high recall of citations among the documents. It is hardly any other document in the set that will generate a similar search key and the risk for two cited references having the same information is also small.
Remeber that you must have an out-file based on \"CR-Cited reference\you must also type in the earliest publication year of the citing documents in the box Min number. When that is done Bibexcel searches the out-file which must be based on cited references. Each time the search key hits the reference lists in the out-file, a pair is formed and saved in a file called \"filename.lin\".This file has the citing document number in the first column and the cited document number in the second.
Thisis what memory.lin will contain:
1 175
2 149
2 236
2 240
2 11
2 59
You may Add labels to units to these pairs in a similar way as to an out-file.When labels to the cited document is to be added you will have to move that column to the left and sort the file ascending to that column (use Excel). However it might be better to use Add labels to docnr-docnr pairs.
This procedure will add data to the pairs in the \"filename.lin\". It is needed in the case that there are several units in the labels (or fields). For example, if thereare several authors in the citing and/or cited document we now treats each authoras a citing or cited author. So, if the citing document has two authors and the cited three authors the citing-cited pair will yield six new pairs.
The result of this procedure will be saved in the file \"filename.add\".
Excluding self-citations
Make an out-file based on the AU-field. Then take the routine: Analyze/Citations among
docs/Add data to links using out-file type.The new file, filename.add, can then be imported
to Excel. In Excel make a new column to the right and use the
Insert/Function logic to test if the citing author is the same as the cited.
You may also go further:
A nice thing about the lin-file is that it can be used to generate co-citations
and shared cited references, thus establishing all three different relations between citing documents in a citation graph. I suggest that you sort the lin-file in Excel taking the A-column as first sort order and the B-column as the second sort order. Use ascending sort order. This means that the form of the lin-file will have the same form as an out file. The lin file will be a reduced out filebased on the cr-field that only contains the document numbers of documentsin the citing document set.
Add labels to docnr-docnr pairs
Bibexcel help
Provided you have made the links you may add labels to each doc in the link. The lin-file (with the citation links) must only have citing-docnr and cited-docnr in sorted order. Otherwise, this will not work! Select the lin-(type)-file and then this routine will read the outfile (must be out-file type) and then add labels from it to the links.
Example
Memory.lin contains citing-cited links among documents, doc nr 1 citing doc nr 175 etc:
1 175
2 149
2 236
2 240
2 11
2 59
if we have memory.out containing authors, select memory.lin and then Citations among docs/Add data to links using out-file type the file memory.add will look like this, each citation producing several citing cited author pairs:
1 BIEHAL GJ 175 BIEHAL G
1 BIEHAL GJ 175 CHAKRAVARTI D
1 GARDIAL SF 175 BIEHAL G
1 GARDIAL SF 175 CHAKRAVARTI D
2 TULVING E 149 CRAIK FIM
2 TULVING E 236 DALLAS M
2 TULVING E 2 TULVING E 2 TULVING E 2 TULVING E 2 TULVING E
236 JACOBY LL
240 MCCLOSKEY M
240 SANTEE J
11 TULVING E
59 WOLTERS G
因篇幅问题不能全部显示，请点此查看更多更全内容