semantics day 38 ling 681.02 computational linguistics harry howard tulane university

18
Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

Upload: susanna-ball

Post on 14-Jan-2016

220 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

SemanticsDay 38

LING 681.02Computational Linguistics

Harry HowardTulane University

Page 2: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

2

Course organization

http://www.tulane.edu/~howard/NLP/

Page 3: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

Analyzing the meaning of sentences

SLPP 10

Page 4: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

4

The problem

Suppose we have a program that lets us type in a natural language question and gives us back the right answer:Which country is Athens in?Greece.

How hard is it to write such a program?

Page 5: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

5

Querying a database

City Country Population

athens greece 1368

bangkok thailand 1178

barcelona spain 1280

berlin east_germany 3481

birmingham united_kingdom 1112

Page 6: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

6

SQL

SQL (Structured Query Language) is a language designed for retrieving and managing data in relational databases.

For example, executing the query below will pull out the value 'greece':SELECT Country FROM city_table WHERE City = 'athens'

Page 7: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

7

Parsing into SQL

Can we get the same effect using English as our input to the query system?

The feature-based grammar formalism described in Chapter 9 makes it easy to translate from English to SQL.

Each phrase structure rule is supplemented with a recipe for constructing a value for the feature sem.

We use the string concatenation operation + to splice the values for the child constituents to make a value for the parent constituent. S[SEM=(?np + WHERE + ?vp)] -> NP[SEM=?np] VP[SEM=?vp]

Page 8: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

8

Philosophical problem

Understanding = looking up in a database?What is a city?What is a country?What is a population?

Page 9: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

9

Depiction of a situation in which Margrietje loves Brunoke

Page 10: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

10

Conclusions

Two fundamental notions in semantics:Declarative sentences are true or false in certain

situations. Definite noun phrases and proper nouns refer to

things in the world.

Page 11: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

11

Consistent vs. inconsistent

5a. Sylvania is to the north of Freedonia.5b. Freedonia is a republic.6a. The capital of Freedonia has a population

of 9,000.6b. No city in Freedonia has a population of

9,000.7a. Sylvania is to the north of Freedonia.7b. Freedonia is to the north of Sylvania.

Page 12: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

12

Logic & NL

Broadly speaking, logic-based approaches to natural language semantics focus on those aspects of natural language which guide our judgments of consistency and inconsistency.

The syntax of a logical language is designed to make these features formally explicit.

As a result, determining properties like consistency can often be reduced to symbolic manipulation, that is, to a task that can be carried out by a computer.

In order to pursue this approach, we first want to develop a technique for representing a possible situation.

Page 13: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

13

Model

We do this in terms of something that logicians call a model.

A model for a set W of sentences is a formal representation of a situation in which all the sentences in W are true.

The usual way of representing models involves set theory.

The domain D of discourse (all the entities we currently care about) is a set of individuals, while relations are treated as sets built up from D.

Page 14: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

14

Example of a model

Our domain D will consist of three children, Stefan, Klaus and Evi = s, k and e. D = {s, k, e}.

Expressions: boy denotes the set

consisting of Stefan and Klaus,

girl denotes the set consisting of Evi,

is running denotes the set consisting of Stefan and Evi.

Page 15: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

15

Propositional logic

Boolean Operator Truth Conditions

negation (it is not the case that ...)

-φ is true in s iff φ is false in s

conjunction (and) (φ & ψ) is true in s iffφ is true in s and ψ is true in s

disjunction (or) (φ | ψ) is true in s iffφ is true in s or ψ is true in s

implication (if ..., then ...)

(φ -> ψ) is true in s iffφ is false in s or ψ is true in s

equivalence (if and only if)

(φ <-> ψ) is true in s iffφ and ψ are both true in s or both false in s

Page 16: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

16

Quiz grades

Q5 Q6 Q7 Q8 Q9 Q10

MIN 7.0 6.5 5.0 7.0 9.0 7.0

AVG 8.8 8.4 8.3 8.8 9.8 7.6

MAX 10.0 10.0 10.0 10.0 10.0 8.0

Page 17: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

30-Nov-2009 LING 681.02, Prof. Howard, Tulane University

17

Missing quiz grades

MichaelHunterLeahCaitlinAlexanderParisa

0P3P1, P5P1P5P6, P9, P10

Page 18: Semantics Day 38 LING 681.02 Computational Linguistics Harry Howard Tulane University

Next time

No quiz

NLPP §10

Analyzing the meaning of sentences