Title: Friendly software should listen and speak
1????????-???????? ???????
Friendly software should listen and speak
- ???????? ??????? ??.3539
- ????? ?????? ??.3539
2????
- ????????
- QA-??????? Start
- ?????????????? ??????? ??? ??????? ?????????????
????? - ?????????????? ????????? ??????
- ??????
3????????
4????????-???????? ??????? ???????????
- ????????-???????? ???????
- ?????????????? ???????
- ?????????
- ??????????
- ????????????????
- ?????????? ???????????-???????? ?????????
5????????? ??????????
- ???? ????????
- ???????? ??????? ? ??????? ????????
- ????????? ?????????
- ????????????? ???????? ????????? ????? ??????????
6??????? ? ??????? ?????????? QA-??????
- 1960-? ??. ?????? QA-???????
- ?????????????? ??? ???????????-???????? ????????
??? ?????????? ?????? - ??????????? QA-???????
- ?????????????????????? (closed-domain)
- ??????? ??????????
- ????????
- ??????????
- ?????????? ??????????
- ??????
-
- ?????????? ? ???????? ???????
- ??????????? ????????? ??? ?????????? ????????
- ????? (open-domain)
- ???????????????
- ???????? ????????? ??????? Start (1993 ?.)
7??????????? QA-???????
8????????
- ???? ????????
- ????????? ????????
- ??????????? ???????
- ????????? ?????? ??? QA-??????
- ????????? ???????
- ???????????? ??????
- ?????? ?? ??????? ? ???????? ???????
- ????????????? ???????
- ???????????????
- ???????? ??????????? (??????)
- ??????? ????????????? QA-??????
9????????-???????? ??????? Start
10QA-??????? Start.
- http//start.csail.mit.edu
- ??????? ? MIT Artificial Intelligence Laboratory
? 1993 ???? ??? ???????????? Boris Katz - ?????????????
- ?????????? ?????????? ????
- ????????? ??????
- ????????? ????????? (???? ??????)
- ???? ????????
- ????????? ??????? Start
- ?????????? ?????? ??????? NLP, ?????????????
?????????? ??? ??????? Start - ???? ???????? ????????? ?? ???????????? ?????? ?
?????? ??????????
11Start. ???? ????????
- ??????? ?? ????????????
- What is a fractal?
- ??????? ? ??????
- Who invented the telegraph?
- ??????? ?? ??????????
- What country is bigger, Russia or USA?
- ????????? ???????
- Show me some poems by Alexander Pushkin
-
12Start. ??????? ????????
- ?????????
- Give me the states that border Colorado.
- What's the largest city in Florida?
- Show me a map of Denmark
- List some large cities in Argentina
- Which is deeper, the Baltic Sea or the North Sea?
- Show the capital of the 2nd largest country in
Asia - ?????????
- When was Beethoven born?
- Who composed the opera Semiramide?
- What movies has Dustin Hoffman been in?
13Start. ??????? ????????
- ????? ? ?????????? ??????????
- What is Jupiter's atmosphere made of?
- Why is the sky blue?
- Convert 100 dollars into Euros
- How is the weather in Boston today?
- How far is Neptune from the sun?
- Show me a metro map of Moscow.
- ??????? ? ????????
- What countries speak Spanish?
- Who was the fifth president of the United States?
- What languages are spoken in the most populous
country in Africa? - How many people live on Earth?
14Start. ???? ??????
doc1
doc2
Parser
Generator
???? ??????
text
doc3
doc4
15Start. ???? ??????
- ??????? ?? 3-? ??????
- ????????? ????????? (T-?????????)
- ??????????????/????????????? ??????? ??????
(S-???????) - ??????? ???? (????????)
16Start. ????????? ?????????
- T-????????? ??? ????????? ????
- lt?????? ????????? ???????gt
- ? ???????? ???????/???????? ?????? T-?????????
????? ????????? ?????? T-????????? - ??????????????, ?????????????? ???????????,
???????? ? ?????? ????? ??????????? ????????????
??? ???????? ?????????????? T-????????? - ????????? ???????? ??????????? (???????, ???????
????????, ???????, ??????????????? ???????, ?????
?????????? ? ?.?.) ???????? ? ???????????
????????? History, ????????? ? T-??????????.
17Start. ?????? T-?????????
Bill surprised Hillary with his answer
Parser
ltltBill surprise Hillarygt with answergt ltanswer
related-to Billgt
18Start. ????????? ???????? ? ??????? T-?????????
Whom did Bill surprise with his answer?
?????????? ????????
Bill surprised whom with his answer?
Parser
ltltBill surprise whomgt with answergt ltanswer
related-to Billgt
???? ??????
Whom Hillary
ltltBill surprise Hillarygt with answergt ltanswer
related-to Billgt
?????????
Bill surprised Hillary with his answer
19Start. ????????? ???????? ? ??????? T-?????????
Did Bill surprise with his answer?
?????????? ????????
Bill surprised Hillary with his answer?
Parser
ltltBill surprise Hillarygt with answergt ltanswer
related-to Billgt
???? ??????
Yes!
?????????
Yes, Bill surprised Hillary with his answer
20?-????????? vs. ???????? ?????
- The bird ate the young snake
- The snake ate the young bird
- The meaning of life
- A meaningful life
- The bank of the river
- The bank near the river
21?-????????? vs. ???????? ?????
- ???????? ?????
- ???????????? ?????????? ? ????????? ?????? ?????
??????? - ?????? ???????????? ?? ?? ?????????????
?????????, ? ?? ?????????????? ???????????????
???????? ???? - T-?????????
- ???????? ??????? ???? ? ??????????? ? ?????????
????? ????? ???? - ????????????? ???? T-????????? ?????????? ???
??????????? ????????? ?? ???????????? ?????? - ?????????? ??? ??????????????
22What do frogs eat?
- ?????, ?????????? ?? T-?????????? ????? 6
???????, ?? ??????? 3 ?????????? - Adult frogs eat mainly insects and other small
animals, including earthworms, minnows, and
spiders - One group of South American frogs feeds mainly on
other frogs - Frogs eat many other animals, including spiders,
flies, and worms
23What do frogs eat?
- ?????, ?????????? ?? ???????? ??????, ????? 33
??????????, ? ??????? ????? ??????????? ?????? ??
?????? What eats frog? ? ?????? ?????????? ????
eat ? frog - Bowfins eat mainly other fish, frogs, and
crayfish - Cranes eat a variety of foods, including frogs,
fishes, birds, and various small mammals.
24Start. S-???????
????????
Bill surprised Hillary with his answer
Bills answer surprised Hillary
Parser
Parser
ltanswer surprise Hillarygt ltanswer related-to Billgt
ltltBill surprise Hillarygt with answergt ltanswer
related-to Billgt
???????
25Start. S-???????
- S-??????? ????????? ??????????????? ????????
- ???????????
- ????????
- ???????????????
- ???????????? ?????
- ??????????????
- ????????
- ????????/????????? ?????
- ?????????????? ??????????????
- ????? ???????????? ??? ???????? ??????????
??????????
26Start. S-???????. ???????
27Start. S-???????
- ????????? S-??????? ????? ?????????????? ? 2-?
???????????? - ? ??????
- ??? ?????????? ???? ?????? ?????? T-???????????
- ? ????????
- ??? ????????? ??????? ????????????
28Start. ????????
- ????????? S-??????? ????????? ? ??????? ????
- ? ????????? ?????????? ???????? ???? ?????,
?????? ??? ??????? ????? ?????? ?????? ?????, ?
??????? ??? ?????????
29WordNet
- ????????????? ???????? ??????????? ?????
- ??????? ?? ???????? (???????)
- ??????
- ????????? ???????????? ????
- ???????? ????????
- ???? ????? ????????? ???????? (????????)
- 150 000 ????, 115 000 ????????, 207 000 ???
????? ??????
30WordNet. ????????????? ????????? ????? ?????????
- ???????????????
- ?????????? Y ????????? X, ???? X
????????????? Y - ???????? Y ??????? X, ???? Y ????????????? X
- ?????? ?? ????? X ? Y ????? ?? ?????, ???? ? ???
????? ????????? - ???????? Y ??????? X, ???? X ????? Y
- ???????? Y ??????? X, ???? Y ????? X
31WordNet. ????????????? ????????? ????? ?????????
- ???????
- ????????? ????????? ??????
- ??????? ??????? ????????
- ????? ?????????? ???????
- ?????? ?????? ?? ????? ??????
32Start. WordNet
- WordNet ???????????? ? ??????? Start ??? ??????
?????????? ? T-??????????? - ????? ? ???? ???? T-????????? ltbird can flygt
- Canary ??????? Bird
- ?? ?????? Can canary fly? Start ??????? Yes
33Start. Omnibase
- ????????????? ????
- ???????????? ??? ?????????? ???????? ? ??????
- ?????? ??????-????????-????????
- ?????? Federico Fellini is a director of La
Strada - ?????? La Strada
- ???????? director
- ???????? Federico Fellini
- ??????? ??????? ??????????? ???????? ?????? (data
source) - Star Wars imdb-movie
34Start. Omnibase. ???????
35Start. Omnibase. ?????? ???????
Who directed gone with the wind?
Start
?????? ???????
Who directed X?, X Gone with the wind
Gone with the wind
Omnibase
????????? ???????
??????? ????????
imdb-movie
(get imdb-movie Gone with the Wind Director)
???????
?????????? ???????
imdb-movie
Victor Fleming
????????? ??????
Victor Fleming directed Gone with the wind
36Start. Omnibase
- ????????????
- ?????? ?????? ???????? ? ????
- ?????????????? ????????????? ??????
??????-????????-???????? - ??????????
- ????????????? ????????? ??????? ??? ???????
????????? ??????
37Start. ?????? ??????? ?????????? ??????
- Wikipedia
- The World Factbook 2006
- Google
- Yahoo
- The Internet Movie Database
- Internet Public Library
- The Poetry Archives
- Biography.com
- Merriam-Webster Dictionary
- WorldBook
- Infoplease.com
- Metropla.net
- Weather.com
38????????????? ???????
- ????? ????????? ???????? ???? ????????
- ???????? ????????? ??????? ??????????,
??????????? ? ???? - ??? ?????????? ? ???? ?????? ??????????? ?? ????
?????? - ????????????
- ????????????
- ??? ???????? ????????????? ???????? ???????
???????????? ?????? RDF (Resource Description
Framework), ?????????? ?? - ?????? XML
- ???????? ?????? ????????? ???????
39Start. Natural Language Annotations
- ???????????? ? ??????? ??????????????? ?????
?????????? ????????? ?? ???????????? ????? - ?????????? ????? ???????-???????? ? ????????????
????????? ?????????? - ? ???? ?????? ???????? ?????? ????????? ?
?????????????? ???????? ?? ????????? - ??????????? ??????????? ??????? ? ??????????
????????????? ???? - ??????
- ???????????
- ???????????
- ???? ??????
- ?????????
- ????????? ????? ???? ???????????????
40Start. Natural Language Annotations
- ????????? ?????????
- ?????????? ????????? ? RDF ???????? ??????????
- ????????????? ????????????????? ????????? (????
??????? ? ??????????) - ????????????? ???? ?????? ???????
41Start. ?????????? ?????????
- How many people live in Kiribati?
- What is the population of the Bahamas?
- Tell me Guams population.
42Start. ?????????? ?????????
- ltrdfsClass ID"Country"gt
- ltrdfscommentgtA Country in the CIA
Factbooklt/rdfscommentgt - lt/rdfsClassgt
- ltrdfProperty ID"population"gt
- ltrdfsdomain rdfresource"Country"/gt
- ltrdfsrange rdfresource"xsdstring"/gt
- ltnlann text"Many people live in ?s"/gt
- ltnlann text"population of ?s"/gt
- ltnlgen text"The population of ?s is ?o"/gt
- lt/rdfPropertygt
43Start. ????????????????? ?????????
- What is the country in Africa with the largest
area? - Tell me what Asian country has the highest
population density. - What country in Europe has the lowest infant
mortality rate? - What is the most populated South American
country?
44Start. ????????????????? ?????????
- ltnlInformationAccessSchemagt
- ltnlanngtwhat country in region has the largest
attributelt/nlanngt - ltnlpatterngt?x a Countrylt/nlpatterngt
- ltnlpatterngt?x map(attribute) ?vallt/nlpatterngt
- ltnlpatterngt?x location regionlt/nlpatterngt
- ltnlactiongtdisplay(boundto(?x,
max(?val)))lt/nlactiongt - ltnlmappinggt
- ltnlhash variable"attribute"gt
- ltnlmap value"population"gtpopulationlt/nlmap
gt - ltnlmap value"area"gtarealt/nlmapgt
- ...
- lt/nlhashgt
- lt/nlmappinggt
- lt/nlInformationAccessSchemagt
45Start. ????????????????? ?????????
- Is Canadas coastline longer than Russias
coastline? - Which country has the larger population, Germany
or Japan? - Is Nigerias population bigger than that of South
Africa?
46Start. ????????????????? ?????????
- ltnlInformationAccessSchemagt
- ltnlanngtcountry-1s att is larger than
country-2s attlt/nlanngt - ltnlpatterngt?x a Countrylt/nlpatterngt
- ltnlpatterngt?x map(att) ?val-1lt/nlpatterngt
- ltnlpatterngt?y a Countrylt/nlpatterngt
- ltnlpatterngt?y map(att) ?val-2lt/nlpatterngt
- ltnlactiongtdisplay(gt(?val-1, ?val-2)))lt/nlactio
ngt - ltnlmappinggt
- ltnlhash variable"att"gt
- ltnlmap value"population"gtpopulationlt/nlmap
gt - ltnlmap value"area"gtarealt/nlmapgt
- ...
- lt/nlhashgt
- lt/nlmappinggt
- lt/nlInformationAccessSchemagt
47Start. ????? ?????? ??????
- What is the distance from Japan to South Korea?
- How far is the United States from Russia?
- Whats the distance between Germany and England?
- ???? ?????? ?? ????? ??????
- ????? ??????? ????? ??????
- ????? ??????? ?????? ??????
- ????????? ?????????? ????? ????
48Start. ????? ?????? ??????
- ltnlInformationPlanningSchemagt
- ltnlanngtdistance between country1 and
country2lt/anngt - ltnlplangt
- ltrdfSeqgt
- ltrdfligtwhat is the capital of country1
?capital1lt/rdfligt - ltrdfligtwhat is the capital of country1
?capital2lt/rdfligt - ltrdfligtwhat is the distance between
?capital1 and ?capital2 - ?distancelt/rdfligt
- lt/rdfSeqgt
- lt/nlplangt
- ltnlactiongtdisplay(?distance)lt/nlactiongt
- lt/nlInformationPlanningSchemagt
49Start. Natural Language Annotations
- ????????????
- ???????? ?????????? ?? ?????????????,
???????????? ????? - ???? ????????????????? ????????? ?????
???????????? ????? ???????? - ????????????? ???????? ??????????? ?????????
- ??????? ? Omnibase ????? ??????????? ? ???????
????????? - ??????????
- ????????? ?????????
50Start. ????? ??????
?????????? ????????
?????? ??????
Parser
??????
???? ??????
T-?????????
WordNet
?????????2
?????
?????????1
?????? ? Omnibase
Omnibase
??????
WWW
doc3
doc2
docN
??????
doc1
passageM
passage1
passage2
?????
????????? ???????
51Start. ??????? ??????????
- ????????? ????????? ???????? ? ?? ??????????????
???????????? - ??????????? ??? ? ????? ????????? ????????????
?????????? - ????????????? ??????? ????????????? ?????? ?
?????????? - ????????? ?????????
- ?????? ????????? ?????? ??????? ??? ?????????
?????????? ????????
52?????????????? ??????? ??? ??????? ?????????????
?????
53??????
54????????
- Salespeople sold the dog biscuits.
55??????????? ????? ????
56??????? ?????? ????? ????
- t ??? (det, noun, )
- w ????? ? ?????? (can, will)
- p(t w) ??????????? ????, ??? ??? t
????????????? ????? w - p(w t) ??????????? ????, ??? ????? w
????????????? ???? t - p(t1 t2) ??????????? ????, ??? t1 ???? ?????
t2
57?????????????
- ????? 90
- ??????????? 97
- ??????? 98
58??????? ?????? ???????
adj large 0.004 small 0.005
0.218
0.45
0.0016
det a 0.245 the 0.586
noun house 0.001 stock 0.001
0.475
59?????? ?????? (transformational tagging)
- ????????? ????? ????????.
- ???? ????? ??????
- ???????? ? ????? ??? X ?? ??? Y, ???? ???
??????????? ????? Z. - ????????? ??? ??????? ???????-?? ???.
- ???????? ???????
- ?????????? HMM vs. ?????????? TT
- (?????????? ????????? ????)
60Treebank
- ?????? ?? ?????? ??????????? ???????, ?????????
????????????? ??????????????? ?????????. - ??????
- (s (np (det The) (noun stranger))
- (vp (verb ate)
- (np (det the) (noun doughnut)
- (pp (prep with) (np (det a) (noun
fork)))))
61??????????? Statistical Parser
- ????????
- ???? ??????? ??????? ?? Pen treebank
- ?????????? ? ????
- ?????????? ?????? ??? ??????????
- ?????????? ???????????? ????????
- ?????????? ???????? ??????????
62PCFG (Probabilistic context-free grammars)
- sp ? np vp (1.0)
- vp ? verb np (0.8)
- vp ? verb np np (0.2)
- np ? det noun (0.5)
- np ? noun (0.3)
- np ? det noun noun (0.15)
- np ? np np (0.05)
63??????? ??????????? ???????????? ??????
- s ????????? ???????????
- ?????????? ???? ??????
- ? ????????? ?????????? ??????? ??????
- r(c) ??????????? ?????????? r ??? ?
64?????????? ??????????? PCFG. ??????? ???????.
- ????? ??????? Pen treebank
- ????????? ?? ???? ??? ???????
- ?????? ?? ??????? ??????
- ????????? ?????? ????? ???????
- P(???????) ?????????? ??? ?????????, ????????
?? ????? ??????????
65Two state-of-the-art statistical parsers. Markov
grammars
- ?????? ???????? ????????????? ????? ?????? ??????
- ???? ?????? ???????? ??????, ???????
??????????? ????, ???, ???????? np prep -
- p(t1 f, t2) ??????????? ????????? ???? t2
????? ???? t1 ?????? ????? f.
66Lexicalized parsing
- ?????? ??????? ?????? ???????? ????? (head),
??????????????? ??. - p(r h) ??????????? ????, ??? ????? ?????????
??????? r ??? ???? ? ???????? h. - p(h m, t) ??????????? ????, ??? ????? h
???????? ???????? ??????? ? head m ? ????? ???
t.
67Lexicalized parsing
- ??????
- (S (NP The (ADJP most troublesome) report)
- (VP may
- (VP be
- (NP (NP the August merchandise trade
deficit) - (ADJP due (ADVP out) (NP tomorrow)))))
- p(h m, t) p(be may, vp)
- p(r h) p(posvp ? aux np be)
68Lexicalized parsing
- the August merchandise trade deficit
- rule np ? det propernoun noun noun noun
69?????????????? ????????? ?????? (Causal
Reconstruction)
70??????
- ???? ?? ?????????? ?????? ???????????? ???????
??????, ?????? ?? ???????????? ? ??????? ????? ?
???????????? ??????? ??????.
71??????
- ?????. ???? ?? ?????????? ?????? ????????????
??????? ??????, ?????? ?? ???????????? ? ???????
????? ? ???????????? ??????? ??????. - ?????????? ?? ??? ??????? ????? ?????
- ?? ???? ??????? ??????? ???????
- ?????? ?? ?????????? ????? 2?
72?????? CR (Causal Reconstruction)
- ???? ????? ???????? ???????? ?? ????????????,
????????? ?????? ????? ???????? ?? ??????????????
???????.
73???????????? ?????????(Transition space)
- ????????????? ????????, ??? ??????? ?????????
- The perception of causality
- ???????????? ?????????
- ??????
- ?????? ???????????? ?? ????????? ????????
74????????? ????????????? (Causal modeling)
75?????? ????????, ????????? ??????? ?????? ??? CR
- ??????????
- ????????
- ?????????
- ?????
76???????? ????????? ??????
77???????? ????????? ??????
- ??????????
- ??????????? ?? ????? ????????
- ????????
- ???????????????
- ????????????
- ?????????
- ???????? ?????????? ????????? ?? ?????????
???????? - ?????
- ????? ?? ????? ????????? ???????
78?????????. 3 ???? ???????????
- ??????????
- ???????? ??????????? ????? ????? ?????????
- ???????????
79?????????????? ????
- ?????????? ????? ??????????? ?????? ? ???????????
- ??????????? ???????
- ?????????????? ???????
- ??????? ??????
- ??????? ?????????????
802 ?????? ????????? ????????? ??????
813 ???? ?????????? ????????.??????? ???????
- ??????? ????????? ? ?????????? ????????? ???????
? ???????? ??????? - ??? ?????????? ? ???????? ??????? ?? ?????
?????????? ?????? Windows?
823 ???? ?????????? ????????.??????? ???????
- ????????? ??????? ????? ?????
- ??? ????????? ????? ????? ????? ?????? ??
????????? ?????? ???????????????? ?????? ? ?????
???????????? - ????????? ????????? ????? ?????????, ?????????? ?
???????, ? ??? ????????? ??????? - ??? ????????? ????????? ????? ???????? ??
???????????
83???????????? ?????????
- ????????????? ???? ????????? ? ??????? ??????? ?
???????? - ?????????? ??????? ??? ??????????????????
breakpoint?? - ???????????? ??????????
- ?????????? ?????????
84????????? ?????? ????????? (States vs. Changes)
- ??????
- Changes ???????????? ? ????
85??????
- The contact between the steam and the metal plate
appears. - The concentration of the solution increases.
- The appearance of the film changes.
- The pin becomes a part of the structure.
- The water remains inside the tank.
86??????
- The contact between the steam and the metal plate
appears. - The concentration of the solution increases.
- The appearance of the film changes.
- The pin becomes a part of the structure.
- The water remains inside the tank.
87????????????? ??????????????
88????????????? ??? NOT-DISAPPEAR
?????????????? ????????
DECREASE
NOT-DECREASE
89??????????? ???????
- ???????? ?????????
- ???????
- ???????-?????????
- ????? ???????
90??????
- APPEAR(contact, ltthe-steam, the-metal-plategt, t1,
t2) - INCREASE(concentration, the-solution, t3, t4)
- CHANGE(appearance, the-film, t5, t6)
- APPEAR(a-part-of, ltthe-pin, the-structuregt, t7,
t8) - NOT-DISAPPEAR(inside, ltthe-water, the-tankgt,
- t9, t10)
91??????????. ?????? ?????
- ltinput sentencegt
- ltattribute-expressiongt ltverb-groupgt
- ltattribute-expressiongt
- the ltattributegt ltprepositiongt ltnoun-phrasegt
- ltprepositiongt and ltnoun-phrasegt
- ltverb-groupgt CHANGE APPEAR ? ?.?.
- The concentration of the solution increases.
92??????????. ?????? ?????
- ltinput sentencegt
- ltobjectgt ltverb-groupgt ltpredicate-modifiergt
- ltpredicate-modifiergt
- ltattributegt ltprepositiongt ltnoun-phrasegt
- ltprepositiongt and ltnoun-phrasegt
- ltverb-groupgt becomes becomes not
remains remains not - The water becomes a vapor.
93??????
- CLEF. http//clef-qa.itc.it/
- WordNet. http//wordnet.princeton.edu/
- Pen treebank. http//www.cis.upenn.edu/treebank/
- Start. http//start.csail.mit.edu/
- TREC. http//trec.nist.gov/
- Eugene Charniak 1997, Statistical Techniques
for Natural Language Parsing - Gary C. Borchardt 1993, Causal Reconstruction
94??????
- Boris Katz, Beth Levin 1988 Exploiting Lexical
Regularities in Designing Natural Language
Systems - Boris Katz and Jimmy Lin. Annotating the Semantic
Web Using Natural Language. September, 2002. - Boris Katz, Sue Felshin, Deniz Yuret, Ali
Ibrahim, Jimmy Lin, Gregory Marton, Alton Jerome
McFarland and Baris Temelkuran. Omnibase Uniform
Access to Heterogeneous Data for Question
Answering. June, 2002. - SEMLP. http//semlp.com/
- RCO. http//www.rco.ru/