Challenges in nlp
-
Upload
zareen -
Category
Data & Analytics
-
view
197 -
download
1
description
Transcript of Challenges in nlp
Ambiguity
• Natural language is highly ambiguous and must be disambiguated.
Ambiguity in Speech
• Speech Recognition – “recognize speech” vs. “wreck a nice beach”
– “youth in Asia” vs. “euthanasia”
1. I saw the man. The man was on the hill. I was using a telescope. 2. I saw the man. I was on the hill. I was using a telescope. 3. I saw the man. The man was on the hill. The hill had a telescope. 4. I saw the man. I was on the hill. The hill had a telescope. 5. I saw the man. The man was on the hill. I saw him using a telescope.
I saw the man on the hill with a telescope.
Ambiguity in Preposition Attachment
Humor and Ambiguity
• Many jokes rely on the ambiguity of language:
– One morning I shot an elephant in my pajamas. How he got into my pajamas, I’ll never know.
– She criticized my apartment, so I knocked her flat.
7
Polysemy Word Sense Disambiguation (WSD)
• Words in natural language usually have a fair number of different possible meanings. – Ellen has a strong interest in computational linguistics. – Ellen pays a large amount of interest on her credit card. – The dog is in the pen. – The ink is in the pen. – I put the plant in the window – Ford put the plant in Mexico
• For many tasks (question answering, translation), the
proper sense of each ambiguous word in a sentence must be determined.
Some more examples of Polysemy
A world record. A record of the conversation. Record it!
He left the bank five minutes ago. He left the bank five years ago He caught a fish at the bank.
I need some paper. I wrote a paper. I read the paper.
Computers are no better than your dog.
But we can teach them “how-to” by coding our knowledge of the language comprehension
process
Co-Reference Resolution
• Determine which phrases in a document refer to the same underlying entity. – John put the carrot on the plate and ate it.
– Bush started the war in Iraq. But the president needed the consent of Congress.
Ellipsis Resolution
• Frequently words and phrases are omitted from sentences when they can be inferred from context.
"Wise men talk because they have something to say; fools, because they have to say something.“ (Plato) "Wise men talk because they have something to say; fools talk because they have to say something.“ (Plato)
16
Information Extraction (IE)
• Identify phrases in language that refer to specific types of entities and relations in text.
• Named entity recognition is task of identifying names of people, places, organizations, etc. in text.
people organizations places – Michael Dell is the CEO of Dell Computer Corporation and
lives in Austin Texas.
• Relation extraction identifies specific relations between entities. – Michael Dell is the CEO of Dell Computer Corporation and
lives in Austin Texas.
Question Answering • Directly answer natural language questions
based on information presented in a corpora of textual documents (e.g. the web). – When was Barack Obama born? (factoid)
• August 4, 1961
– Who was president when Barack Obama was born? • John F. Kennedy
– How many presidents have there been since Barack Obama was born? • 9
Projects & Research
Wikitology: A Novel Hybrid Knowledge Base
Derived from Wikipedia
Zareen Syed
Ou r plan s a re going ah ead , He ylig hen i s
get ting ticke ts, so let's put
th at in in ha rd pencil, but keep yo ur eraser h and y a ll the same. My life is
re ally hectic, and I won 't be in a sane
pla ce ti ll mid-April . We 'll p indow n the details later, but a t this po int
we we re consi dering Val onl y to
ta lk on "The Metasystem Tra nsition as th e Quantum of Evolu tion". Th is is
th e theoretical ba se to the PCP, which I
described the form of in my ta lkto WESS. I t's b asically be tw een Fran cis
and Val ho w would like to talk, or
both, or what.
Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's put
th at in in hard pen cil, but keep your erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane
pla ce ti ll mid -April. We 'll p indown th e detai ls later, b ut at this point
we we re consideri ng Va l only to
ta lk on "The Metasystem Transitio n as th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I
described the form of in my talkto WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, or
both, or wha t.
Ou r plan s a re going ah ead , He ylig hen i s get ting ticke ts, so let's put
th at in in ha rd pencil, but keep yo ur
eraser h and y a ll the same. My life isre ally hectic, and I won 't be in a sane
pla ce ti ll mid-April . We 'll p in
dow n the details later, but a t this po int we we re consi dering Val onl y to
ta lk on "The Metasystem Tra nsition as
th e Quantum of Evolu tion". Th is isth e theoretical ba se to the PCP, which I
described the form of in my ta lk
to WESS. I t's b asically be tw een Fran cis and Val ho w would like to talk, or
both, or what.
Our plans are go ing a head, H eyli ghen is
ge tt in g tickets, so let's p ut
that in in h ard pe ncil, bu t keep your era ser handy all the sa me. My li fe is
rea lly hectic, an d I wo n't be in a sane
place till mid-Apri l. We'll pindo wn the de tails late r, but at this p oin t
we were con side ring Val o nly to
talk on "The Me tasyste m Transition a s the Quan tu m o f Evol ution". This is
the theo retical b ase to th e PCP, which I
de scribed th e fo rm o f in my talkto WESS. It 's basically betwe en Franci s
an d Va l how would like to talk, or
bo th , or wh at.
Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's putth at in in hard pen cil, but keep your
erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane pla ce ti ll mid -April. We 'll p in
down th e detai ls later, b ut at this point
we we re consideri ng Va l only tota lk on "The Metasystem Transitio n as
th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I described the form of in my talk
to WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, orboth, or wha t.
Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's put
th at in in hard pen cil, but keep your erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane
pla ce ti ll mid -April. We 'll p indown th e detai ls later, b ut at this point
we we re consideri ng Va l only to
ta lk on "The Metasystem Transitio n as th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I
described the form of in my talkto WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, or
both, or wha t.Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's put
th at in in hard pen cil, but keep your erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane
pla ce ti ll mid -April. We 'll p indown th e detai ls later, b ut at this point
we we re consideri ng Va l only to
ta lk on "The Metasystem Transitio n as th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I
described the form of in my talkto WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, or
both, or wha t.
Ou r plan s a re going
ahe ad, Heyl igh en is
get ting ticke ts, so let's put
th at in in ha rd
pen cil, but keep your era ser handy
all the same. My life
Ou r plan s a re going ah ead , He ylig hen i s
get ting ticke ts, so let's putth at in in ha rd pencil, but keep yo ur
eraser h and y a ll the same. My life is
re ally hectic, and I won 't be in a sane pla ce ti ll mid-April . We 'll p in
dow n the details later, but a t this po int
we we re consi dering Val onl y
19
Introduction and Motivation
Page 20
Ou r plan s a re going ah ead , He ylig hen i s
get ting ticke ts, so let's put
th at in in ha rd pencil, but keep yo ur eraser h and y a ll the same. My life is
re ally hectic, and I won 't be in a sane
pla ce ti ll mid-April . We 'll p indow n the details later, but a t this po int
we we re consi dering Val onl y to
ta lk on "The Metasystem Tra nsition as th e Quantum of Evolu tion". Th is is
th e theoretical ba se to the PCP, which I
described the form of in my ta lkto WESS. I t's b asically be tw een Fran cis
and Val ho w would like to talk, or
both, or what.
Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's put
th at in in hard pen cil, but keep your erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane
pla ce ti ll mid -April. We 'll p indown th e detai ls later, b ut at this point
we we re consideri ng Va l only to
ta lk on "The Metasystem Transitio n as th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I
described the form of in my talkto WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, or
both, or wha t.
Ou r plan s a re going ah ead , He ylig hen i s get ting ticke ts, so let's put
th at in in ha rd pencil, but keep yo ur
eraser h and y a ll the same. My life isre ally hectic, and I won 't be in a sane
pla ce ti ll mid-April . We 'll p in
dow n the details later, but a t this po int we we re consi dering Val onl y to
ta lk on "The Metasystem Tra nsition as
th e Quantum of Evolu tion". Th is isth e theoretical ba se to the PCP, which I
described the form of in my ta lk
to WESS. I t's b asically be tw een Fran cis and Val ho w would like to talk, or
both, or what.
Our plans are go ing a head, H eyli ghen is
ge tt in g tickets, so let's p ut
that in in h ard pe ncil, bu t keep your era ser handy all the sa me. My li fe is
rea lly hectic, an d I wo n't be in a sane
place till mid-Apri l. We'll pindo wn the de tails late r, but at this p oin t
we were con side ring Val o nly to
talk on "The Me tasyste m Transition a s the Quan tu m o f Evol ution". This is
the theo retical b ase to th e PCP, which I
de scribed th e fo rm o f in my talkto WESS. It 's basically betwe en Franci s
an d Va l how would like to talk, or
bo th , or wh at.
Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's putth at in in hard pen cil, but keep your
erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane pla ce ti ll mid -April. We 'll p in
down th e detai ls later, b ut at this point
we we re consideri ng Va l only tota lk on "The Metasystem Transitio n as
th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I described the form of in my talk
to WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, orboth, or wha t.
Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's put
th at in in hard pen cil, but keep your erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane
pla ce ti ll mid -April. We 'll p indown th e detai ls later, b ut at this point
we we re consideri ng Va l only to
ta lk on "The Metasystem Transitio n as th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I
described the form of in my talkto WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, or
both, or wha t.Ou r p lan s are going ahead , Heylighen is
getting t icke ts, so let 's put
th at in in hard pen cil, but keep your erase r hand y all the same . My life is
re ally hect ic, and I won't be i n a sane
pla ce ti ll mid -April. We 'll p indown th e detai ls later, b ut at this point
we we re consideri ng Va l only to
ta lk on "The Metasystem Transitio n as th e Qu antum of Evolution". Th is is
th e th eore ti cal base to the PCP, wh ich I
described the form of in my talkto WESS. It's basica lly be tween Fran cis
and Val how wou ld li ke to tal k, or
both, or wha t.
Ou r plan s a re going
ahe ad, Heyl igh en is
get ting ticke ts, so let's put
th at in in ha rd
pen cil, but keep your era ser handy
all the same. My life
Ou r plan s a re going ah ead , He ylig hen i s
get ting ticke ts, so let's putth at in in ha rd pencil, but keep yo ur
eraser h and y a ll the same. My life is
re ally hectic, and I won 't be in a sane pla ce ti ll mid-April . We 'll p in
dow n the details later, but a t this po int
we we re consi dering Val onl y
Human mind capable of understanding and reasoning over knowledge in different forms and is influenced by contextual factors
World Knowledge is available in different forms
Context important in understanding the semantics of data And may be available in different forms
Related Work
Page 21
Michael Jackson
Michael Joseph Jackson (August 29, 1958 – June 25, 2009) was an American singer-songwriter, dancer, actor, choreographer, businessman, philanthropist and record producer.
Contents Life and Career Death . . See Also References External Links
Typical Wikipedia Article
Linked Open Data
Wikipedia Derived Knowledge Resources
Support Structured
Queries
Supports Natural
Language Queries
Wikitology
Supports Hybrid
Queries
Wikitology • Linked to LOD Cloud with over 295 datasets
22
Wikitology Document Concept Prediction
• Identifying the topics and concepts associated with a document or collection of documents is a common task for many applications such as:
– Annotation and categorization of documents in a corpus
– Modelling user interests
– Business intelligence
– Selecting Advertisements
Page 23
Test Document Title Method 1 Ranking Categories Directly
Method 2 Spreading Activation
Pulses=3
Weather Prediction of thunder storms (CNN) “Weather_Hazards” “Meterology”
Prediction for Single Test Document
Experiments
More pulses -> More Generalized Concepts
Data Set Method 1 Ranking Categories
Directly
Method 2 (2 pulses) Spreading Activation on
Category links Graph
Method 3 (2 pulses) Spreading Activation on
Article Links Graph
10 articles related to Organic Farming
Agriculture (Rank 1) Agriculture (in Top 5) Organic_farming (Rank 1)
Prediction for a Set of Documents
Concept not in the
Category Hierarchy
Wikitology Cross Document Co-reference Resolution
• Problem:
– Determine whether various named entities in different documents refer to the same object in the world. • Are two documents that talk about “George Bush” talking about the same
George Bush?
– defined as a task in ACE
Page 25
Wikitology Entity Linking
• Research Problem: – Given an entity mention string and an article with that
entity mention, find the link to the right Wikipedia entity if one exists.
– Defined as a task in TAC KBP Track
Page 26
John Williams
Richard Kaufman goes a long way
back with John Williams. Trained as a
classical violinist, Californian Kaufman
started doing session work in the
Hollywood studios in the 1970s. One of
his movies was Jaws, with Williams
conducting his score in recording
sessions in 1975...
John Williams author 1922-1994
J. Lloyd Williams botanist 1854-1945
John Williams politician 1955-
John J. Williams US Senator 1904-1988
John Williams Archbishop 1582-1650
John Williams composer 1932-
Jonathan Williams poet 1929-
Knowledge Base
Identify matching entry, or determine that entity is missing from KB
Automatic Discovery of Slots and Fillers
Page 27
Slot Score Fillers Example
Musician 1.00 ray_charles, sam_cooke ...
Album 0.99 bad_(album), ...
Location 0.97 gary,_indiana, chicago, …
Music_genre 0.90 pop_music, soul_music, ...
Label 0.79 a&m_records, epic_records, ...
Phonograph_
record 0.67
give_in_to_me, this_place_hotel
…
Act 0.59 singing
Movie 0.46 moonwalker …
Company 0.43 war_child_(charity), …
Actor 0.41 stan_winston, eddie_murphy,
Singer 0.40 britney_spears, …
Magazine 0.29 entertainment_weekly,…
Writing_style 0.27 hip_hop_music
Group 0.21 'n_sync, RIAA
Song 0.20 d.s._(song) …
New Slots Album Movie Phonograph_record/songs Musician (related Musicians) Act
Wikitology Architecture and API
Page 28
A Broader Unified Framework for Automatically Enriching Wikitology
Page 29
CONCEPT
PREDICTION
INFORMATION
EXTRACTION
PART OF SPEECH
TAGGING
CLUSTERING
CLASSIFICATION
SENTIMENT
ANALYSIS TAXONOMY
MANAGEMENT
ENTITY LINKS
GRAPH
Atomic_bombings_of_Hiroshima_and_Nagasaki Enola_Gay George_Weller Little_Boy
"Sixteen hours ago an American airplane dropped one bomb on Hiroshima, Japan, and destroyed its usefulness to the enemy. That bomb had more power than 20,000 tons of T.N.T. It had more than two thousand times the blast power of the British Grand Slam, which is the largest bomb ever yet used in the history of warfare".These fateful words of the President on August 6th, 1945, marked the first public announcement of the greatest scientific achievement in history. The atomic bomb, first tested in New Mexico on July 16, 1945, had just been used against a military target.On August 6th, 1945, at 8:15 A.M., Japanese time, a B-29 heavy bomber flying at high altitude dropped the first atomic bomb on Hiroshima. More than 4 square miles of the city were instantly and completely devastated. 66,000 people were killed, and 69,000 injured.On August 9th, three days later, at 11:02 A.M., another B-29 dropped the second bomb on the industrial section of the city of Nagasaki, totally destroying 1 1/2 square miles of the city, killing 39,000 persons, and injuring 25,000 more.On August 10, the day after the atomic bombing of Nagasaki, the Japanese government requested that it be permitted to surrender under the terms of the Potsdam declaration of July 26th which it had previously ignored.
Title
Enola Gay was the name of the aircraft
Weller's reports from Nagasaki after the nuclear bombing were censored by the United States military but appeared in a book in 2002.
"Little Boy" was the codename of the atomic bomb dropped on Hiroshima
Predicted Concepts
None of this information is present as
words in the given text!
Little Boy – Keyword Search
Keyword search retrieves irrelevant
documents in results as well
32
Example: 1
Query : Little Boy
More than 100,000 Results
Field : wikiconceptref
Query : Little Boy
A conceptual search only retrieves relevant articles related to
the “little boy” concept
100,000 results vs.
26 Relevant Results
BotColony
Botcolony
20Q game
https://www.botcolony.com/ppSD2/custom/get_started/register-trial.php
BotColony 3D Game
Thank you
Wikitology Related Publications
1. Z. Syed and T. Finin. "Creating and Exploiting a Hybrid Knowledge Base for Linked Data", LNCS, Springer-Verlag. 2010. (submitted)
2. Z. Syed and T. Finin. “Approaches for Enriching Wikipedia”. In Proc. of the AAAI-2010 Workshop on Collaboratively-built Knowledge Sources and Artificial Intelligence. 2010.
3. Z. Syed and T. Finin. “Unsupervised techniques for discovering Ontology elements from Wikipedia article links”. International Workshop on Formalisms and Methodology for Learning by Reading (FAM-LbR). 2010.
4. Z. Syed, T. Finin and V. Mulwad. “Exploiting a Web of Semantic Data for Interpreting Tables”. In Proc. of Web Science Conference, WebSci’2010.
5. T. Finin and Z. Syed. "Creating and Exploiting a Web of Semantic Data", In Proc. of the Second International Conference on Agents and Artificial Intelligence. Jan. 2010.
6. T. Finin, Z. Syed, J. Mayfield, P. McNamee, and C. Piatko, "Using Wikitology for Cross-Document Entity Coreference Resolution", Proceedings of the AAAI Spring Symposium on Learning by Reading and Learning to Read, March 2009.
Page 39
Wikitology Related Publications
7. J. Mayfield, D. Alexander, B. Dorr, J. Eisner, T. Elsayed, T. Finin, C. Fink, M. Freedman, N. Garera, P. McNamee, S. Mohammad, D. Oard, C. Piatko, A. Sayeed, Z. Syed, R. Weischedel, “Cross-Document Coreference Resolution: A Key Technology for Learning by Reading”, AAAI 2009 Spring Symposium on Learning by Reading and Learning to Read, March 2009.
8. Z. Syed and T. Finin. "Wikitology: A Novel Hybrid Knowledge Base derived from Wikipedia", In Proc. of the Grace Hopper Celebration of Women in Computing Conference, October 2009. (Abstract)
9. Z. Syed and T. Finin. "Wikitology: Wikipedia as an ontology", In Proc. of the Grace Hopper Celebration of Women in Computing Conference, October 2008. (Abstract)
10. Z. Syed, T. Finin and A. Joshi. 2008. “Wikipedia as an Ontology for Describing Documents”. In Proc. of the International Conference on Weblogs and Social Media. 2008.
11. Mark Dredze, Paul McNamee, Delip Rao, Adam Gerber, and Tim Finin, “Entity Disambiguation for Knowledge Base Population”, Proceedings of the 23rd International Conference on Computational Linguistic. 2010.
Page 40