Abstract 
A computer program for automatic identification of "full- 
form" case citations in legal literature (e.g., Rutherford v. Geddes, 
4 Wall. 220, 18 L. Ed. 343; Southland Industries, Incorporated v. 
Federal Cc~mmlnications Ccm~nlssion, 1938, 69 App. D.C., 82, 99 Fo 
2D 117) has been developed at the University of Pittsburgh and is 
now operational. 
The level of performance of this program known as "The 
Citation Identifier" is high. In a recent computer run, ~ae 
Citation Identifier scanned the full texts of 191 randomly select- 
ed decisions of U.S. Court of Appeals (some 400, 000 words of run- 
ning text) and located correctly 2,220 full-form citations out of 
a total of 2, 227 (that is better than 99% of the total). Only 
seven misses and three false drops occurred. 
Of 2, 220 full-form citations which were located correct- 
ly, 1944 (87%) were identified perfectly. In addition, there 
were 276 partial identifications containing two types of errors : 
(1) partial identifications in which some citation terms were 
mistakenly lopped off by the program (so-called "short hits"), and 
(2) partial identifications which contained words that were im- 
properly included in the citations (so-called "long hits"). 
Both types of errors are for the most part easily cor- 
rectable and can be largely eliminated by suitable changes in the 
program. 
-1- 
The Citation Identifier operates rather rapidly. In a 
recent test run, the total time required to process some 400, 000 
running words of t~xt was approximately fifteen and a half 
minutes. This speed could be further increased by suitable 
changes in the cc~uter program. 
An extension of the Citation Identifier to reduced-form 
citations (e.g., "The Geddes decision," "the Southland Industries 
ease") is no~ in preparation. 
Motivation for Automatic Identification of Case Citations in Legal 
Literature. 
Efficient administration of justice has often been 
hampered by the slowness ~nd inefficiency of ordinary methods of 
legal information h~dling. 
Because lawyers, judges, government attorneys, legisla- 
tors, and others find access to necessary legal data often slow and 
inefficient, they are frequently unable to act with the speed and 
the effectiveness which the circumstances may demand of them. Thus, 
for example, members of the legal profession are often unable to 
procure, promptly, exhaustive and accurate data concerning legal 
precedents of various court decisions. This has often hampered the 
initiation of legal actions, slOWed down the preparation of de- 
fense and offense, delayed the preparation of new laWs, and other- 
Wise interfered with the efficiency of legal processes. 
However, wise automation of certain critical areas of 
legal information processing could alleviate considerably the 
present crisis in legal documentation. For example, a set of 
-2- 
comber progrs~s capable of exhaustive, accurate, rapid, and 
economical identificatio~ of legal precedents in legal literature 
would do ~ch to eliminate fr~ legal documentation one of its 
most serious bottlenecks. 
Since many or most legal precedents are referred to in 
legal literature by means of full-form and reduced-form case ci- 
tatiens, (e. g., respectively, '~ealy v. Penna. RR. Co., supra" 
and '~he Healy case"), antc~atic identification of both forms of 
case citations would go far in the direction of automatic identi- 
fication of leg~l precedents in legal literature. 
However, a perusal of legal texts shows that automatic 
identification of full-form citations is both ~ch simpler than 
that of reduced-form citations and also a prerequisite for effi- 
cient identification of the latter. 
Therefore, construction of a set of programs for auto- 
rustic identification of legal precedents in legal literature has 
begun with the construction of a computer progrma for automatic 
identification in legal texts of full-fc~m legal citations. 
State of the Art and the Genesis of The Citation Identifier. 
Practical work in legal aut~tie information retrieval 
has until now revolved mainly around: (I) the preparation of 
co~cords~ces to legal te~ts, (2) EWIC indexin 8 of legal texts, 
and (3) matching of legal texts with the key terms of queries and 
inte~eet profiles. These well-establishe~ automatic information 
retrieval activities have met with considerable success. 1'2 
"3- • 
In additioa, a sizable ~mount of time and energy has 
been devoted to automatic construction frc~ computer-readable 
legal texts of legal indexes and legal thesauri by means of 
various statistical techniques. 3' 4 However, to the best of our 
knowledge, these interesting procedures have not been incorporated 
into any practical legal information systems. 
No previous attempts at autcm~tic identification of 
case names in legal texts have c~ne to our attention; however, 
this general type of activity has been extensively discussed by 
Casimir Borkowsi and derives directly from his efforts sdmed at 
automatic identification in texts of classes of words and of word 
Strings referring to various types of individuals, objects, pro- 
cesses, acts, relaticas, groups, etc. 5, 6, 7, 8, 9 
The Citation Identifier was first undertaken by 
Borkowski and his students in the Department of Cc~puter Science 
of the University of Pittsburgh, in early 1968 as part of a work- 
shop section of a graduate course in automatic text processing. 
Research and development procedures adapted by Borkowski and his 
group were approximately as follows : 
A set of challenging but nevertheless resolvBble problems 
in automatic text processing w~s presented to the class, discussed, 
and resolved in a general way. A detailed solution of one of the 
probl~ns, namely, automatic identification of full-form case 
J 
citations, w~s then worked out, flowcharted and programmed in a 
high-level progra~ language (PENELOPE) I0 for the IBM 360/50 
computer of the University of Pittsburgh. 
Computer-readable legal texts were made available to 
the class by Aspen Systems Corporation (formerly, the Health 
Law Center of the University of Pittsburgh)~ and both the 
teacher and the students want to take this opportunity to thank 
Aspen Systems for these data. 
Since the termination of the original classroom work- 
shop project, The Citation Identifier was reprogrsamed in 0S/360 
assembly lauguage for Aspen's IBM 360/40 by Sperling Martin s one 
of the students, who is also with Aspen Systems Corporation. 
The effectiveness of this recent version of The 
Citation Identifier constitutes, along with an outline of its 
present structure, the subject matter of this paper. 
Structure of The Citation Identifier 
Our rules for automatic identification of citations are 
essentially simple. Full-form case citations in legal texts are 
recognized by means of s straightforward identification procedure 
whose main steps are listed below: 
1. Copy a sentence from a computer-readable document into an 
area in computer memory (hereafter, "the Search Ares"). Then 2 
starting at the beginning of the Search Area. 
2. Search the text from left to right for an occurrence of 
"v. ". NOTE: The presence of "v. " (for "versus") within a sen- 
tence is taken to indicate the presence in that sentence of a 
full-form case citation. 
3. (A) If "v." is not found, return to 1. above for next 
instruction. 
-5- 
(B) If "v." is found, record its Search Area location 
and go to #. below for next instruction. 
4. Starting at the location of "v. " search the sentence from 
right to left for the first occurrence of a string of characters 
which m~tches either : 
(A) A string of characters which is on a list of so-called 
"Left Delimiters of a Case Citation" (hereafter, "LD") 
(see NOTE i. below), 
or else 
(B) A string of characters which is on a llst of so-called 
'potential Left Delimiters of a Case Citation" (hereafter, 
'~LD") (see NOTE 2. below)° 
NOTE l: Entries on LD list (of which there are approximstely 
one hundred) are: (a) words such as: "also", "although", "cites", 
"cited", "in", "note", "see" 3 "since", "when", etc., (b) abbrevia- 
tions such as: "c°f.", "e.g.", "ViZo" etc. 3 and (c) punctuation 
m~rks such as: colon, semicolon, sentence period (i.e. a period 
followed by two or more spaces), a question m~rk, etc. 
NOTE 2: The only two entries on PLD llst are the word "of" and 
the ccmm~ punctuation mark. 
5. (A) If the character string to the left of "v." was mstch- 
ed by an entry on LD llst, then: 
(a) flag as the beginning of tke case citation the 
first word to the right of that character string, 
(b) return to the location of "v." in the Search Area, 
4- 
(c) go to 9- belov far the next instruction. 
I~0TE: The first occurrence of a string of characters to the 
left of '~." which m~tches an entry on Left Delimiters List is 
interpreted as the first element outside the case citation. In 
other words, we seek to locate the first occurrence of a string 
of characters Vhich is net within the well-formed formula for 
the "left me~aber" of a i~O_l-fo~m case citatio~ or -- to put it 
yet another way -- we are on the lookout for the first string 
of characters which is in the cc~Ixlement of the set of all formu- 
las for the left ~ers of c~se citations. 
(B) If the chezacte~ string to the left of "v." was match- 
ed by an entry on P~D list, go to 6. below. 
6. (A) If the chs~acter string in the sentence was matched by 
the entry "of" on PT~ list, then (a) note its location in the 
sentence, and (b) check whether the text word which is to its 
immediate left is m~tched by an entry on the list of so-called 
"Resolvers i" (see ~ i. below) and go to 7. below for the 
next instruction. 
(B) If the ~cter in the sentence was m~tched by the 
entry "," (i.e. a c~) on PLD list, then check whether 
the text word which is to its immediate right is on the 
list of so-called '~esolvers 2" (see NOTE 2. below) and go 
to 8. below for the next instruction. 
ROTE i: List of Resolvers i contains words such as "authority", 
"citation", "l~w", "rea~onlng", "rule", etc. 
2: List of Resolvers 2 contains words such as "Incorporated" 
-7- 
and "Limited", abbreviations such as "Inc. "3 and "Ltd. ", and 
abbreviations of names of states : "Ariz. "3 "Ark. "3 "cal. ", etc. 
7. If the string of characters in text was matched by an entry 
• on List of Resolvers 13 then: 
(A) (a) flag as the beginning of case citation the first 
word to the right of the word "of" 3 
(b) return to the location of "v° " in the Search Area, 
(c) go to 9. below for the next instruction; 
otherwise 
(B) Starting at the location of the ccsmm, continue exe- 
cuting the instruction 4. above. 
8. (A) If the string of characters in text Was not matched 
by an entry on List of Resolvers 2, then: 
(a) flag as the beginning of case citation the first 
word to the right of the comma, 
(b) return to the location of "v. " in the Search Area, 
(c) go to 9. below for the next instruction; 
otherwise 
(B) Starting at the location of the contain 3 continue exe- 
cuting the instruction 4. above. 
9- Starting at the location of "v. " search the sentence from 
left to right for either: 
(A) The first occurrence of a sentence period 3 
or 
(B) A string of characters which is a number, 
or 
-8- 
W 
(C) A string of characters which matches an entry on the 
list of so-called '~Bibliography Terms" (hereafter, '~T") 
(see NOTE i. below). 
NOTE i: Entries on BT list (of Which there are approximately 
one hundred ~nd fifty) are: 
(A) Words sad phrases such as: "affirmed" 3 "ante~ "at 
p84~e" 3 "certoriemi denied" 3 "certoriari granted" 3 "Docket" 3 
"infra" 3 "super" 3 "supra" 3 etc. 
(B) Abbreviations such as: "aff'd" 3 "A.L.R.", "app." 3 
"Atl. ", "A. 2d" 3 "Cranch. "3 "Cir. "~ "C. C. "3 '7. Supp. "3 etc., 
and the names of states referred to in 6. above. 
NOTE 2: During this part of the program~ an aut(~nation is made 
to scan the "right member" of the citation to check its well- 
formedness. During this scan, we are on the lookout for all 
strings which are in the set of well-formed formulas for right 
members of full-form citations. The right boundary marker is 
pl~ced to the left of the first character string which is believed 
not to be part of the well-formed formula for the right member. 
i0. (A) If the sentence period is encountered, then 
(a) flag as the end of the citation the string of 
characters to its immediate left, 
(b) print the citation and go to 1. above for the 
next instruction. 
(B) If a string of characters in the text was either 
identified as a number or else was matched by an entry on 
BT list, then go to ll. below for the next instruction. 
-9- 
ll. Continue searching the text from left to right for either 
the first occurrence of the sentence period or a string of 
characters which matches an entry on BT list. 
12. (A) If a sentence period is encountered, go to lO. (A) 
(a) above for the next instruction. 
(B) if the string of characters in the text was either 
identified as a number or was matched by an entry on BT 
list, then continue executing ll. above; 
otherwise 
(a) flag as the last element of the case citation~ 
the word to the left of that chs~acter string, 
(b) print the citation, 
(c) remain at present location in the Search Area and 
go to 2. above for the next instruction. 
Effectiveness of The Citation Identifier 
The assembly language version of The Citation Identifier 
as reported here was developed and tested out on Aspen Syst~ns 
Corporation's ISM 360/40. Legal texts used in the test were the 
decisions of the United States Court of Appeals (Third Circuit). 
In a recent computer run, The Citation Identifier scan- 
ned 191 randomly selected court decisions (45, 942 lines of texts, 
that is some 400,000 words of running text) and.located correct- 
ly 2, 220 full-form citations out of a total of 2, 227 (that is 
better than 99% of the total). 
Of 2, 220 full-form citations which were located correct- 
ly, l~ 9~J4 (87%) were identified Perfectly. In addition, there 
-lO- 
I 
I 
were 276 partial identifications containing two types of errors : 
i. "Short hits"~ i.e. full-form case citations in which some 
citation terms were mists~enly lopped off by the program~ The 
ns~ber of such l~rtial identifications was 208 (e.g., "Carlino 
v. Zimblarte~ 15~7, 60" for "Carlino v. Zimblarte, 1927, 60 
Ontario Law ~e~a~s 269"), 
and 
2. "Long hits", i.e. i~Lll-form case citations which contained 
terms which were imprOl~rly included in the citations. The 
number of such partially c~rect identifications was 82 (e.g., 
"Gulf v. Schlumberger ~ is an ordinary civil action"). 
The Citation Identifier operates rather rapidly. The 
total time required to process over 400, 000 running words of 
text and print out 2~220 full-form citations was appraximately 
fifteen and one half minutes. 
-ii- 
TABLE I 
Results of the Experiment 
It~n: Number : 
Full-text documents in the sample ........................ 191 
Running words of texts ........................ over 400, 000 
Full-form case citations in the sample .................. 2, 227 
Full-form case citations located correctly .............. 2,220 
Full-form case citations missed ............................. 7 
False drops ................................................. 3 
Perfect identifications ................................. l, 944 
Partial identifications ................................... 276* 
Short hits ................................................ 208 
Long hits .................................................. 82 
Job time (in minutes) ........................ under 16 
Of 276 partial Identifications~ * 31 were caused by 
typographic errors; 18 and l~O, respectively, were due to lack 
of appropriate entries on the lists of Left Delimiters and 
Bibliographic Terms, and 9 and 12, respectively, to lack of en- 
tries on Resolvers 1 and Resolvers 2 lists. The misses, false 
drops, and the remainder of partial identifications (apprc~imate- 
ly 75) were caused by various incorrect assumptions incorporated 
I 
* Because some partial hits contained both short and long hits s 
the total of short hits and long hits is greater th~n the number 
of partial identifications. 
into the basic identlfic~tion routines. Thus~ for example, the 
assumption that all right members of full-form citations terminate 
in strings of numbers and bibliographic terms has caused seven 
misses 8~1d nineteen long hits (e.g., the following string of 
words was identified as a single citation, '~cCullough v. 
Cosgrave cited its previous opinion in Los Angeles Brush Corp. v. 
James, 1927, 272 U.S. 701"). 
Similarly~ the assumption that the presence of "v. " 
in a sentence indicates the presence in that sentence of a full- 
form case citation resulted in false drops such as: "C. L. 
McClain Fuel Corp. v. appellsmt's contention that the case at 
bar fe/_Is within this testimony". 
Sone changes in the structure of the main identifica- 
tion procedure are noW in l~relma-ati~. A preliminary evaluation 
indicated that they should lead to further significant improve- 
ments in the accuracy and the speed of the program. A preliminary 
evaluation indicates that by increasing the number of list entries 
by about a factor of four, we would reduce the number of partial 
hits by about a factor of three. Among required neW entries ~fnich 
will be added to the list are: frequently mispelled words and 
abbreviatlons~ left delimiters, bibliographic terms, etc. 
We would estim~te that the introduction into The Citation 
Identifier of all modifications suggested above would reduce by 
a factor of five or six the number of partial identifications. 
--13- 
J 
Discussion and Inter~pretation 
We would like to emphaize the fact that the task which 
We set for ourselves was the solution of a ~ practical 
~oblem. Consequently, we did not think it appropriate to commit 
ourselves to any strong theoretical view of ordinary language and 
sought instead to discover what minimal assumptions and what in- 
formation may be pertinent to our unprepossessing experiments in 
automatic text processing. 
Simple~ hypotheses Concerning automatic identification 
in texts of case citations were selected by us with the intention 
of finding out how many correct identifications and how many 
errors they would produce. It was and it r~nains our plan to 
amend these hypotheses on a continuous basis in the light of the 
results obtained. We are, of course, striving for stronger 
theoretical underpinnings; however, for the time being, we find 
it appropriate to operate with the least s~ount of preconceived 
opinion and of theoretical commitment. 
Automatic classification of words and phrases in texts 
of the type described here can be viewed as a particularly simple 
case of machine translation fr~n ordinary language. However, the 
goal of The Citation Identifier is not translation into natural 
language but into classificatory language. In other words, our 
program attempts a relatively simple many-tp-one type of reduction 
(i. e. classification) rather than the extremely ccmplex many-to- 
many transformation of the '~T" type. 
More generally , it may be useful to view ordinary 
language as a macro-language containing certain special-purpose 
mlcro-languages (or '~ini-lsaguages") -- each with its own struc- 
ture which relative to the total structure of language is quite 
simple. It may be of considerable practical and theoretical 
interest (a) to investigate the structures and the interrela- 
tions of such mini-languages and (b) to construct computer pro- 
grams for identifying in te~ts the words and the word strings 
belonging to such mini-languages. 
An ability to produce and identify autc~atlcslly words 
and word strings belonging to various speclal-purpose categories 
(i. e. mini-languages, each with its own set of gre~n,atlcal rules) 
should be very useful in information retrieval because they play 
an important role in various systems for extracting and dlstribut ~- 
ing information. 
Because many word strings which the algorithms such as 
this one attempt to identify have simple structure ("phrase: 
structure"), they can be recognized with a reasonable degree of 
accuracy by means of simple ccmputatlonal techniques. 
The Citation Identifier is the first of a series of 
programs for automatic and semiautcmatlc processing of cenputer- 
readable legal texts. An extension of The Citation Identifier 
to reduced-form citations (e.g., "the Geddes decision", "the 
Geddes ease") is nOW in preparation. In addition, The Citation 
Analyzer~ a c~nputer program for automatic classification of 
fuSl-form case citations is also in preparation. 
Several uses suggest th~nselves for a caaputer program 
capable of identifying cheaply, rapidly, accurately, and exhaus- 
tively case citations in legal literature. They seem to fall 
into five broad and overlappin~ categ~ies : 
i. Aut~n~tic indexing and classification of legal literature, 
2. Establishing counts of occurrences of case citations in 
case law and in statutory law, 
3- Determining ho~ case citations co-occur with other words 
and phrases in legal texts (this m~y lead eventually to corre- 
lating case citations with points-of-law), 
4. Tracin~ associations between case citations and construction 
of lists, tables, and graphs which display such associations, 
5. Providing an automatic or s~miautQm~tic service for answer- 
i~ questions concerning documents in which a particular case or 
group of cases was cited. 
Syst~As for aut(m~tic identification of citations in 
texts and subsequent autc~tic extraction of case citations fr~n 
texts may be useful to many groups, among the: 
i. Lawyers, judges, government attorneys, and other members 
of the legal profession, 
~. M~nbers of various legislatures, 
3. Officials in v~rious branches of the federal, state, and 
m~uicipal governments, 
4. Administrators in business, industry, foundations, labor# 
finance, insurance, transportation, etc., 
5. Sociologists sz~ l~olitical scientists, 
and many others. 
-18- 
FIG. I 
NO 
® 
~ES 
ELEME 
I MARK LEFT I " BOUNDAB.Y OF 
CITATION 
POITNoTER NO 
SET POINTER 
" TO RIGHT 
I SET POINTER I rO NEXT RIGHT 
CHARACTER 
STRING. . 
FLOWCHART OF THE CITATION IDENTIFIER 
-19- 
179 F.2d 695 (3rd Cir. 1950} National Labor Relations Board v. Spiewak 
National Labor Relations Board v. Spiewak et al. ! in No. 987~T 
Counsel, Arnold Ordman, Washington, D.C. (David P. Findling, Associate 
den. Counsel, &. Norman Somers, Asst. Gun. Counsel, Marcel Ballet-Prevost, 
Washington, D.C. , on the brief} , for petitioner. 
Counsel, Gerald R. Chambers, New Yrok City (Chambers and Chambers, New York 
City, on the brief} , for respondents. 
• Sitting, BIGGS, Chief Justice, and MABIS, GOODRICH, NcLAUGRLIN, O-CONRELI, 
K&LODNEN and HASTIE, Gircuit Judges. 
RcLAUGHLIN, Circuit Judge. 
This is 8 petition by the National Labor Relations Board for enforcement of 
its order against respondents following procee6ings under Section 10 of the 
National Labor Relations Act, ~9 Star. 4~9, U.S.C.A. Title 29, Sec. 160. 
Respondents are garment maD-fac~,rers. During the period with W~ich we are 
the Association prior to the wildcat ~trike. There is a pattern indicated by 
the named incidents that takes them out of the category of unimportant casual 
conversation between the individuals concerned, as illustrated by Ouaker ~o 
Oil Refining Corp. v~ N.L.R.B. r 3 Cir. r 119 F.2d 631 r 633 I or the type of 
episodes outlined in N.L.R.B.v. Public Service Co-ordinated Transport et al. 
3 Cir. , 177 F.2d 119, which obviously had no effect in either preventing or ~** 
from membership, even where suc action had been based on the employee's dual 
~ionism. coluate-Palmolive-Peet Co. v. N~L.R.B. . 338 U.S. ~55. 70 ~.c+. 
~66. Nut the conduct of NewfieJd and Nlein bore no relationship to that type .*** 
under Section 8 (1) of the Act. It renderea the 194~ contract invalid. Labor 
Board v. Electric Cleaner Co. f 315 U.S. 685f 69~f 62 S.Ct. 846f 86 L.Ed. 
.~ Cf. Wallace CORD. V. Labor Board. 323 U.S. 2Q~. 6~ S.Ct. 238. A9 L. Ed. 
216... 
though respondents had the contract right to discharge employees 
participatine in the illegal ~rike, (N.~.R.B. v. Sands flf~. Co. f 306 U.S. 
332 r 59 S.Ct. 508 r 83 L.~d. 682: N.L.R.B.v. Fansteel corp. r 306 U.S. 240. 
2U9r 59 S.Ct. ~90 r 83 L.Ed 627: Ratter of Scullin Steel Co. , 65 N.L.R.B. 129a; 
Ratter of Joseph D~son and Sons, 72 N.L.~.B. ~5) , Spiewak's subsequent 
testimoDv points out that in his above answer be was not reCerrinq to such 
employees. It is this which di ti gushes the instant situation from the 
Stackpole and Republic cases ~.L.R.B.v. Stackpole Carbon Co. r 3 Cir. r I0~ 
F.2d 167; Republic Steel corporation v. N.L.R.B. r 3 Cir. r 107 F.2d ~7~) where 
the employer was denied the right to withhold reinstatement of employees who 
had particinated in ~inor acts of violence in furtherance of a strike. 
resuit in the light of the particular facts of this phase of the matter. 
N.L.R.B. v. Wytheville Knitting ~illst 3 Cir. r 175 F.2d 238, 2~0, 
presented a somewhat similar situation, though in that case there was no 
.department into confusion. Nat on I Labor Relations Rd. v. Edinhurg Citrus 
Ass-n. 5 Cir. . lg~5, I~7 F.2d 353. In such circumstances thes e employees 
within*th~ * We do not think it comes language of Section 10 (e} of the Act 
reading: " No objection * * * not * * * urged before ~e ~ard, its member, 
agent, or agency, shall be considered by the COUPT * * * " (See N.L.R.B.v. 
theme 7 California Lamber Co. , 327 U.S. 385, at paues 388 and 389 r 66 S.Ct. 
~53. 90 L.Ed. 739; May Department Stores Co. v. N.E.R.B. ~ 326 U.S. 376, 
foo%not~ at Daoe 386 r 66 $~$. 203~ 209 r 90 i. Ed. I~5; Marshall Field and Co. 
• v. N.L.R.B. ~ 318 U.S. 253 r 255 r 63 S.Ct. 585~ 87 L.Ed. 7~I; N.L.R.B.v. 
Baldwin Locomotive Works r 3 Cir. # 128 F.2d 3 t at page 50) Eveu if it were 
Judged to be under that section, the exception to 10 {e) would apply. That 
exception reads: " * * * unless the failure or neglect to urge such objection 
G. 2 TEXT PORTIONS OF A DOCUMENT CONTAINING FULL-FORM CASE CITATIONS (UNDERLINED) 
~CESSED BY THE CITATION IDENTIFIER---SEE THE RESULTS OF PROCESSING IN FIG. 3. 
-Z0- 
o 
~4 
O 
O ,,4 
r~ 
Ol 
r- 
o 
o 
o 
,,~ ® o 
.ca irl 
• • ta~ 
4a I:= 
• . ,el 
~ ~ • 
o ~J • . 0 
c~ 
ea • . 
0 
t~ U 
ca ~ o ~.~- 
-Z1- 
i' 
I-I 
M 
r,Q 
' H 
~ 0 HU 
a~ 
H 
H i 
ME.~ " 
¢,,.1H 
1"t 
;E.Im 
.I.4M 
O ~M 
1.41-1 ~U 

References 

1. Hoz~y, J. F., Retz~_eval of Statutory and Case ~ 1965 
c~puter Law ~itute, wa~h~on, D.C. (1965). 

2. LITE, General System Description~ Staff Judge Advocate, Air 
Force Accounting and Finance Center, Denver (1967). 

3. Dennis, S. F., "The Design and Testing of a Fully Automatic 
Ind~xlng-Searching System for Documents Consisting of 
Erpository Texts" in Schecter, G. (Editor) : Information 
Retrieval: A Critical View; Thompson, Washington, D. C. (1967). 

4. Kayton, I., "Retrieving Case Law by Computer: Fact, Fiction, 
and Future", George Washington Law Revie~ 35 (No. i): 1-49 
(1%6). 

5. Borkowski, C., A System for Autc~Btic Recognition of Personal 
Ns~es in NeWsl~per Text~ Report RC-1563, Watson IBM Research 
Center, Yorktown Heights, N. Y. (1966). 

6. Borkowski, C., "An Experimental System for Automatic 
Identification of personal Nsm~s and Personal Titles in Newspaper 
Te~ts", American Docgmentation~ 18 (No. 3): 131-138 (1967). 

7. Borkowski, C., "An Experimental System for Autnmatic 
Recognition of Personal Names in Newspaper Texts", Deuxieme 
Conference International sur le TTaitement Autcmatique des 
Laugues (Proceedings), (Section 25: 1-15), Grenoble (1967). 

8. Borkowski, C., "Sc~e Principles and Techniques of Automatic 
Assignment of Words and Word Strings in Texts to Special-Purpose 
Sublanguages", in Gerbner, G. (Editor): The Analysis of 
Ccmmmnieation Content; Scientific Approaches and Computer 
~A~lications~ Wiley, New York (in press). 

9. Borkewski, C., "Syntax and Semantics of Lunese I, a Micro- 
language for Labeling Topographical Features of the Lunar Surface", 
Tenth International Congress of 0nusBstic Sciences (Proceedings), 
Vienna (September 1969). 

I0. Isner, D. W. and B. J. Erllc~ P~ELOPE Reference Manual~ 
Computer Center, University of Pittsburgh (in press). 
