Information RetrievalSearch Engine Technology
(3)http://tangra.si.umich.edu/clair/ir09
Prof. Dragomir R. Radev
SET/IR – W/S 2009
…5. Evaluation of IR systems Reference collections TREC…
Relevance
• Difficult to change: fuzzy, inconsistent
• Methods: exhaustive, sampling, pooling, search-based
Contingency table
w=tp x=fn
y=fp z=tn
n2 = w + y
n1 = w + x
N
relevant
not relevant
retrieved not retrieved
Precision and Recall
Recall:
Precision:
w
w+y
w+x
w
Exercise
Go to Google (www.google.com) and search for documents on Tolkien’s “Lord of the Rings”. Try different ways of phrasing the query: e.g., Tolkien, “JRR Tolkien”, +”JRR Tolkien” +Lord of the Rings”, etc. For each query, compute the precision (P) based on the first 10 documents returned by AltaVista.
Note! Before starting the exercise, have a clear idea of what a relevant document for your query should look like. Try different information needs.
Later, try different queries.
n Doc. no Relevant? Recall Precision1 588 x 0.2 1.00
2 589 x 0.4 1.00
3 576 0.4 0.67
4 590 x 0.6 0.75
5 986 0.6 0.60
6 592 x 0.8 0.67
7 984 0.8 0.57
8 988 0.8 0.50
9 578 0.8 0.44
10 985 0.8 0.40
11 103 0.8 0.36
12 591 0.8 0.33
13 772 x 1.0 0.38
14 990 1.0 0.36
[From Salton’s book]
P/R graph
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Recall
Pre
cis
ion
P/R graph
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Recall
Pre
cis
ion
Interpolated average precision (e.g., 11pt)Interpolation – what is precision at recall=0.5?
Issues
• Why not use accuracy A=(w+z)/N?
• Average precision
• Average P at given “document cutoff values”
• Report when P=R
• F measure: F=(2+1)PR/(2P+R)
• F1 measure: F1 = 2/(1/R+1/P) : harmonic mean of P and R
Kappa
• N: number of items (index i)
• n: number of categories (index j)
• k: number of annotators
)(1
)()(
EP
EPAP
N
i
n
jij k
mkNk
AP1 1
2
1
1
)1(
1)(
2
1
1
)(
Nk
mEP
N
iijn
j
Kappa example
J1+ J1- TOTAL
J2+ 300 10 310
J2- 20 70 90
TOTAL 320 80 400
Kappa (cont’d)
• P(A) = 370/400 = 0.925• P (-) = (10+20+70+70)/800 = 0.2125• P (+) = (10+20+300+300)/800 = 0.7875• P (E) = 0.2125 * 0.2125 + 0.7875 * 0.7875
= 0.665• K = (0.925-0.665)/(1-0.665) = 0.776• Kappa higher than 0.67 is tentatively
acceptable; higher than 0.8 is good
Sample TREC query<top><num> Number: 305<title> Most Dangerous Vehicles
<desc> Description: Which are the most crashworthy, and least crashworthy, passenger vehicles? <narr> Narrative: A relevant document will contain information on the crashworthiness of a given vehicle or vehicles that can be used to draw a comparison with other vehicles. The document will have to describe/compare vehicles, not drivers. For instance, it should be expected that vehicles preferred by 16-25 year-olds would be involved in more crashes, because that age group is involved in more crashes. I would view number of fatalities per 100 crashes to be more revealing of a vehicle's crashworthiness than the number of crashes per 100,000 miles, for example.</top>
LA031689-0177FT922-1008LA090190-0126LA101190-0218LA082690-0158LA112590-0109FT944-136LA020590-0119FT944-5300LA052190-0048LA051689-0139FT944-9371LA032390-0172
LA042790-0172LA021790-0136LA092289-0167LA111189-0013LA120189-0179LA020490-0021LA122989-0063LA091389-0119LA072189-0048FT944-15615LA091589-0101LA021289-0208
<DOCNO> LA031689-0177 </DOCNO><DOCID> 31701 </DOCID><DATE><P>March 16, 1989, Thursday, Home Edition </P></DATE><SECTION><P>Business; Part 4; Page 1; Column 5; Financial Desk </P></SECTION><LENGTH><P>586 words </P></LENGTH><HEADLINE><P>AGENCY TO LAUNCH STUDY OF FORD BRONCO II AFTER HIGH RATE OF ROLL-OVER ACCIDENTS </P></HEADLINE><BYLINE><P>By LINDA WILLIAMS, Times Staff Writer </P></BYLINE><TEXT><P>The federal government's highway safety watchdog said Wednesday that the Ford Bronco II appears to be involved in more fatal roll-overaccidents than other vehicles in its class and that it will seek to determine if the vehicle itself contributes to the accidents. </P><P>The decision to do an engineering analysis of the Ford Motor Co. utility-sport vehicle grew out of a federal accident study of theSuzuki Samurai, said Tim Hurd, a spokesman for the National Highway Traffic Safety Administration. NHTSA looked at Samurai accidents afterConsumer Reports magazine charged that the vehicle had basic design flaws. </P><P>Several Fatalities </P><P>However, the accident study showed that the "Ford Bronco II appears to have a higher number of single-vehicle, first event roll-overs,particularly those involving fatalities," Hurd said. The engineering analysis of the Bronco, the second of three levels of investigationconducted by NHTSA, will cover the 1984-1989 Bronco II models, the agency said. </P><P>According to a Fatal Accident Reporting System study included in the September report on the Samurai, 43 Bronco II single-vehicleroll-overs caused fatalities, or 19 of every 100,000 vehicles. There were eight Samurai fatal roll-overs, or 6 per 100,000; 13 involvingthe Chevrolet S10 Blazers or GMC Jimmy, or 6 per 100,000, and six fatal Jeep Cherokee roll-overs, for 2.5 per 100,000. After theaccident report, NHTSA declined to investigate the Samurai. </P>...</TEXT><GRAPHIC><P> Photo, The Ford Bronco II "appears to have a highernumber of single-vehicle, first event roll-overs," a federal officialsaid. </P></GRAPHIC><SUBJECT><P>TRAFFIC ACCIDENTS; FORD MOTOR CORP; NATIONAL HIGHWAY TRAFFIC SAFETY ADMINISTRATION; VEHICLE INSPECTIONS;RECREATIONAL VEHICLES; SUZUKI MOTOR CO; AUTOMOBILE SAFETY </P></SUBJECT></DOC>
TREC (cont’d)
• http://trec.nist.gov/tracks.html• http://trec.nist.gov/presentations/presentations.ht
ml
Most used reference collections
• Generic retrieval: OHSUMED, CRANFIELD, CACM
• Text classification: Reuters, 20newsgroups• Question answering: TREC-QA• Web: DOTGOV, wt100g• Blogs: Buzzmetrics datasets• TREC ad hoc collections, 2-6 GB• TREC Web collections, 2-100GB
Comparing two systems
• Comparing A and B
• One query?
• Average performance?
• Need: A to consistently outperform B
[this slide: courtesy James Allan]
The sign test
• Example 1:– A > B (12 times)– A = B (25 times)– A < B (3 times)– p < 0.035 (significant at the 5% level)
• Example 2:– A > B (18 times)– A < B (9 times)– p < 0.122 (not significant at the 5% level)– http://www.fon.hum.uva.nl/Service/Statistics/Sign_Tes
t.html [this slide: courtesy James Allan]
The t-test
• Takes into account the actual performances, not just which system is better
• http://www.socialresearchmethods.net/kb/stat_t.php
SET/IR – S/W 2009
…6. Automated indexing/labeling Compression…
Indexing methods
• Manual: e.g., Library of Congress subject headings, MeSH
• Automatic: e.g., TF*IDF based
LOC subject headings
http://www.loc.gov/catdir/cpso/lcco/lcco.html
A -- GENERAL WORKSB -- PHILOSOPHY. PSYCHOLOGY. RELIGIONC -- AUXILIARY SCIENCES OF HISTORYD -- HISTORY (GENERAL) AND HISTORY OF EUROPEE -- HISTORY: AMERICAF -- HISTORY: AMERICAG -- GEOGRAPHY. ANTHROPOLOGY. RECREATIONH -- SOCIAL SCIENCESJ -- POLITICAL SCIENCEK -- LAWL -- EDUCATIONM -- MUSIC AND BOOKS ON MUSICN -- FINE ARTSP -- LANGUAGE AND LITERATUREQ -- SCIENCER -- MEDICINES -- AGRICULTURET -- TECHNOLOGYU -- MILITARY SCIENCEV -- NAVAL SCIENCEZ -- BIBLIOGRAPHY. LIBRARY SCIENCE. INFORMATION RESOURCES (GENERAL)
MedicineCLASS R - MEDICINESubclass RR5-920 Medicine (General)R5-130.5 General worksR131-687 History of medicine. Medical expeditionsR690-697 Medicine as a profession. PhysiciansR702-703 Medicine and the humanities. Medicine and disease in relation to
history, literature, etc.R711-713.97 DirectoriesR722-722.32 Missionary medicine. Medical missionariesR723-726 Medical philosophy. Medical ethicsR726.5-726.8 Medicine and disease in relation to psychology. Terminal care. DyingR727-727.5 Medical personnel and the public. Physician and the publicR728-733 Practice of medicine. Medical practice economicsR735-854 Medical education. Medical schools. ResearchR855-855.5 Medical technologyR856-857 Biomedical engineering. Electronics. InstrumentationR858-859.7 Computer applications to medicine. Medical informaticsR864 Medical recordsR895-920 Medical physics. Medical radiology. Nuclear medicine
Automatic methods
• TF*IDF: pick terms with the highest TF*IDF scores
• Centroid-based: pick terms that appear in the centroid with high scores
• The maximal marginal relevance principle (MMR)
• Related to summarization, snippet generation
Compression
• Methods– Fixed length codes– Huffman coding– Ziv-Lempel codes
Fixed length codes
• Binary representations– ASCII– Representational power (2k symbols where k
is the number of bits)
Variable length codes• Alphabet:
A .- N -. 0 -----B -... O --- 1 .----C -.-. P .--. 2 ..---D -.. Q --.- 3 ...—E . R .-. 4 ....-F ..-. S ... 5 .....G --. T - 6 -....H .... U ..- 7 --...I .. V ...- 8 ---..J .--- W .-- 9 ----.K -.- X -..-L .-.. Y -.—M -- Z --..
• Demo:– http://www.scphillips.com/morse/
Most frequent letters in English
• Most frequent letters:– E T A O I N S H R D L U
• Demo:– http://www.amstat.org/publications/jse/secure/v7n2/
count-char.cfm • Also: bigrams:
– TH HE IN ER AN RE ND AT ON NT
Huffman coding
• Developed by David Huffman (1952)• Average of 5 bits per character (37.5%
compression)• Based on frequency distributions of
symbols• Algorithm: iteratively build a tree of
symbols starting with the two least frequent symbols
Symbol Frequency
A 7
B 4
C 10
D 5
E 2
F 11
G 15
H 3
I 7
J 8
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
c
b d
f
g
i j
he
a
Symbol Code
A 0110
B 0010
C 000
D 0011
E 01110
F 010
G 10
H 01111
I 110
J 111
Exercise
• Consider the bit string: 01101101111000100110001110100111000110101101011101
• Use the Huffman code from the example to decode it.
• Try inserting, deleting, and switching some bits at random locations and try decoding.
Extensions
• Word-based
• Domain/genre dependent models
Ziv-Lempel coding
• Two types - one is known as LZ77 (used in GZIP)
• Code: set of triples <a,b,c>
• a: how far back in the decoded text to look for the upcoming text segment
• b: how many characters to copy
• c: new character to add to complete segment
• <0,0,p> p• <0,0,e> pe• <0,0,t> pet• <2,1,r> peter• <0,0,_> peter_• <6,1,i> peter_pi• <8,2,r> peter_piper• <6,3,c> peter_piper_pic• <0,0,k> peter_piper_pick• <7,1,d> peter_piper_picked• <7,1,a> peter_piper_picked_a• <9,2,e> peter_piper_picked_a_pe• <9,2,_> peter_piper_picked_a_peck_• <0,0,o> peter_piper_picked_a_peck_o• <0,0,f> peter_piper_picked_a_peck_of• <17,5,l> peter_piper_picked_a_peck_of_pickl• <12,1,d> peter_piper_picked_a_peck_of_pickled• <16,3,p> peter_piper_picked_a_peck_of_pickled_pep• <3,2,r> peter_piper_picked_a_peck_of_pickled_pepper• <0,0,s> peter_piper_picked_a_peck_of_pickled_peppers
Links on text compression
• Data compression:– http://www.data-compression.info/
• Calgary corpus:– http://links.uwaterloo.ca/calgary.corpus.html
• Huffman coding:– http://www.compressconsult.com/huffman/ – http://en.wikipedia.org/wiki/Huffman_coding
• LZ– http://en.wikipedia.org/wiki/LZ77
100 alternative search engines
• http://rss.slashdot.org/~r/Slashdot/slashdot/~3/83468703/article.pl
Readings
• 2: MRS9
• 3: MRS13, MRS14
• 4: MRS15, MRS16