Natural Language Processing Lecture 19 11 5 2013 Jim Martin Today Wrap up dependency parsing again Semantics meaning The next couple of lectures are a mix of stuff from Chapters 17 and 18 11 14 13 Speech and Language Processing Jurafsky and Martin 2 1 Four Problems Training Getting training data is tricky Our treebanks associate sentences with their corresponding trees But what we need are parser states paired with their corresponding correct operators That s not in the treebank But we do know the correct trees So 11 14 13 Speech and Language Processing Jurafsky and Martin 3 Four Problems Acquiring Training Data During this phase we ll parse with our standard algorithm asking an oracle which operator to use at any given time During this phase the oracle has access to the correct tree for this sentence At each stage it chooses as a case statement 1 Left if the resulting relation is in the correct tree 2 Right if the resulting relation is in the correct tree AND if all the other outgoing relations associated with the word are already in the relation list 3 Otherwise Shift 11 14 13 Speech and Language Processing Jurafsky and Martin 4 2 Four Problems Training The result of the previous phase is training data not training We ll have a training corpus consisting of search states paired with operator actions That s the data we ll use to train our classifier With features extracted from search states And L R S operators as gold standard labels 11 14 13 Speech and Language Processing Jurafsky and Martin 5 Break Quiz 1 and Quiz 2 Quiz 2 moved to Nov 19 I ll post revised readings Quiz 1 average was 44 50 11 14 13 Speech and Language Processing Jurafsky and Martin 6 3 Transition First we did words morphology Then simple sequences of words Then we looked at syntax Now we re moving on to meaning Where some would say we should have started to begin with 11 14 13 Speech and Language Processing Jurafsky and Martin 7 Example Even if this is the right tree what does that tell us about the meaning 11 14 13 Speech and Language Processing Jurafsky and Martin 8 4 Meaning Representations We re going to take the same basic approach to meaning that we took to syntax and morphology We re going to create representations of linguistic inputs that capture the meanings of those inputs But unlike parse trees these representations aren t primarily descriptions of the structure of the inputs 11 14 13 Speech and Language Processing Jurafsky and Martin 9 Meaning Representations In most cases they are simultaneously representations of the meanings of utterances and representations of some potential state of affairs in some world 11 14 13 Speech and Language Processing Jurafsky and Martin 10 5 Meaning Representations What does this mean representations of linguistic inputs that capture the meanings of those inputs For us it means Representations that permit or facilitate semantic processing 11 14 13 Speech and Language Processing Jurafsky and Martin 11 Semantic Processing Ok so what does that mean Representations that Permit us to reason about their truth i e their relationship to some world Permit us to answer questions based on their content Permit us to perform inference answer questions and determine the truth of things we don t already know to be true 11 14 13 Speech and Language Processing Jurafsky and Martin 12 6 Semantic Processing Touchstone application is often question answering Can a machine answer questions involving the meaning of some text or discourse What kind of representations do we need to automate that process 11 14 13 Speech and Language Processing Jurafsky and Martin 13 Q A 11 14 13 Speech and Language Processing Jurafsky and Martin 14 7 Q A William Wilkinson s An Account of the Principalities of Wallachia and Moldavia inspired this author s most famous novel 11 14 13 Speech and Language Processing Jurafsky and Martin 15 Semantic Processing We re going to discuss 2 ways to attack this problem just as we did with parsing There s the principled theoretically motivated approach Computational Compositional Semantics Chapters 17 and 18 And there are limited practical approaches that have some hope of actually being useful Information extraction Chapter 22 11 14 13 Speech and Language Processing Jurafsky and Martin 16 8 Semantic Analysis Compositional Analysis Create a FOL representation that accounts for all the entities roles and relations present in a sentence Similar to our approach to full parsing Information Extraction Do a superficial analysis that pulls out only the entities relations and roles that are of interest to the consuming application Similar to chunking 11 14 13 Speech and Language Processing Jurafsky and Martin 17 Information Extraction preview Investigators worked leads Monday in Riverside County where the car was reported stolen and reviewed security tape from Highway 241 where it was abandoned said city of Anaheim spokesman John Nicoletti 11 14 13 Speech and Language Processing Jurafsky and Martin 18 9 Information Extraction Named Entities Investigators worked leads Monday in Riverside County where the car was reported stolen and reviewed security tape from Highway 241 where it was abandoned said city of Anaheim spokesman John Nicoletti Investigators worked leads Monday in Riverside County where the car was reported stolen and reviewed security tape from Highway 241 where it was abandoned said city of Anaheim spokesman John Nicoletti 11 14 13 Speech and Language Processing Jurafsky and Martin 19 Information Extraction Events Investigators worked leads Monday in Riverside County where the car was reported stolen and reviewed security tape from Highway 241 where it was abandoned said city of Anaheim spokesman John Nicoletti Investigators worked leads Monday in Riverside County where the car was reported stolen and reviewed security tape from Highway 241 where it was abandoned said city of Anaheim spokesman John Nicoletti 11 14 13 Speech and Language Processing Jurafsky and Martin 20 10 Representational Schemes We re going to make use of First Order Logic FOL as our representational framework Not because we think it s ideal Many of the alternatives turn out to be either too limiting or They turn out to be notational variants 11 14 13 Speech and Language Processing Jurafsky and Martin 21 FOL Allows for The analysis of truth conditions Allows us to answer yes no questions Supports the use of variables Allows us to answer questions through the use of variable binding Supports
View Full Document
Unlocking...