Anthology Statistics

Jun 27 2008
From Drago Radev, ACL Secretary

Here are some statistics that I had promised to send out. They show, for papers in the Anthology, the numbers of citations to other papers by publication type (P=ACL, W=WS, J=CLJ, A=ANLP, N=NAACL, E=EACL, H=HLT, I=IJCNLP).

Here is how to read this table. For example, all ACL05 papers combined included a total of 849 citations to other Anthology papers. Of these, 512 were to other conference papers, 191 were to workshop papers, and 143 were to (CL) journal papers.

I haven't (yet) calculated the impact factors of each conference and the journal using the Anthology data. However, one set of statistics is worth pointing out, namely that 75% of all citations in the journal to other Anthology papers go to conference and workshop papers and that 85% of all citations in ACL proceedings go to conference and workshop papers. In other words, a typical ACL/CL paper cites 4-5 times as many conference/ws papers than journal papers.

My conclusion, as stated at the exec meeting, is that in terms of advancing our field, conference and workshop papers are quite significant and we should treat them as such. In the sense of impact, our conference papers are indeed comparable to journal papers.

I expect objections, e.g., looking at the average number of citations per conference vs. journal paper N years down the road, etc. I will be happy to compute such statistics if needed. Also, the Anthology only includes the CL journal and not JNLE, etc. but again it doesn't include AAAI and SIGIR either.

From	ACL03	ACL04	ACL05	ACL06	ACL07	J03	J04	J05
(total)	528	664	849	2054	1318	262	230	272
P	162	220	281	663	456	77	76	89
W	117	144	191	465	313	48	39	53
J	104	135	143	287	163	59	70	55
C	71	54	69	222	93	32	22	26
A	33	26	45	70	31	18	9	14
N	13	50	90	147	134	4	5	14
E	12	16	16	55	49	7	3	7
H	9	15	11	119	61	15	6	9
I	0	0	0	21	9	0	0	0
CONF	300	381	512	1297	833	153	121	159
WS	117	144	191	465	313	48	39	53
JRNL	104	135	143	287	163	59	70	55