360 likes | 473 Views
עיבוד שפות טבעיות - שיעור אחד עשר Unification (cont.) Semantics. עידו דגן המחלקה למדעי המחשב אוניברסיטת בר אילן. Feature structures in the grammar. CF grammar rules can be augmented with feature structures and with unification operations to express constraints on the constituents of a rule
E N D
עיבוד שפות טבעיות - שיעור אחד עשרUnification (cont.)Semantics עידו דגן המחלקה למדעי המחשב אוניברסיטת בר אילן 88-680
Feature structures in the grammar • CF grammar rules can be augmented with feature structures and with unification operations to express constraints on the constituents of a rule • An example notation (the PATR-II formalism):β 0β 1... β n {set of constraints} • Where the constraints have one of the following two forms: • < βi feature path> =(unify) atomic value • < βi feature path> =(unify) < βj feature path> • e.g.S NP VP<NP NUMBER> = <VP NUMBER> 88-680
Feature structures in the grammar • S NP VP{NP AGREEMENT} = {VP AGREEMENT} • This flight serves breakfast • These flights serve breakfast • S Aux NP VP{Aux AGREEMENT} = {NP AGREEMENT} • Does this flight serve breakfast? • Do these flights serve breakfast? 88-680
Feature structures in the grammar • NP Det Nominal<Det AGREEMENT> = <Nominal AGREEMENT><NP AGREEMENT> = <Nominal AGREEMENT> • this flight vs. these flights 88-680
Feature structures in the grammar • Lexical constituents receive their agreement features directly from the lexicon • Aux does<Aux AGREEMENT NUMBER> = sg<Aux AGREEMENT PERSON> = 3 • Det this<Aux AGREEMENT NUMBER> = sg • Det these<Aux AGREEMENT NUMBER> = pl 88-680
Feature structures in the grammar • Verb serve<Verb AGREEMENT NUMBER> = pl • Verb serves<Verb AGREEMENT NUMBER> = sg<Verb AGREEMENT PERSON> = 3 • Non-lexical constituents(e.g. VPs) receive agreement values from their constituents • VP Verb NP<VP AGREEMENT> = <Verb AGREEMENT> 88-680
Feature structures in the grammar • Agreement (NP and Nominal) • Noun flight<Noun AGREEMENT NUMBER> = sg • Noun flights<Noun AGREEMENT NUMBER> = pl • Nominal Noun<Nominal AGREEMENT> = <Noun AGREEMENT> 88-680
Feature structures in the grammar • For most grammatical categories, the features are copied from one child to the parent • The child that provides the features is called the head of the phrase (the features are the head features) • VP Verb NP<VP AGREEMENT> = <Verb AGREEMENT> • NP Det Nominal<Det AGREEMENT> = <Nominal AGREEMENT><NP AGREEMENT> = <Nominal AGREEMENT> • Nominal Noun<Nominal AGREEMENT> = <Noun AGREEMENT> 88-680
Subcategorization • VP Verb {VP SUBCAT} = INTRANS • VP Verb NP {VP SUBCAT} = TRANS • VP Verb NP NP {VP SUBCAT} = DITRANS 88-680
Semantics 88-680
Representing Meaning • Up to now, we have not focussed at all on what things mean. • How does one represent meaning? • meaning representations:representations that bridge that gap from linguistic forms to knowledge of the world. 88-680
Meaning Representations • What can serve as a meaning representation... • Anything that serves the core practical purposes of a program that is doing semantic processing: • Answering questions • Determining truth • Making inferences 88-680
semantic analysis • הפעולה של השמת מבני משמעות לרכיבים לשוניים נקראת ניתוח סמנטי semantic analysis 88-680
The architecture Sentence Syntactic Structure Parsing Parse Tree Semantic Analysis Semantic Representation 88-680
Applications • Ideally, applications would interact with the semantic representation of texts. • Need a representation which is unambiguous, even though the text is ambiguous. 88-680
Problem of Ambiguity • אין התאמה חח“ע בין משפט ובין הייצוג הסמנטי שלו (רב משמעות) וגם לא להפך: 88-680
Canonical Form • Inputs that mean the same thing should have the same meaning representation! • ייצוג משמעות אחיד לקלטים שונים נקרא canonical form • מאפשר פישוט של תהליך ההסקה, אם המערכת צריכה להתמודד רק עם ייצוג יחיד. 88-680
Canonical Form • John gave a car to Mary • Mary was given a car by John • give(John, Mary, a car) • ∃x ∃y : give(x) ∧ agent(x, John) ∧ recipient(x,Mary) ∧ object(x,y) ∧ car(y) 88-680
Example • Every bird lives on a tree • Quantifier and word sense ambiguities. • ∀x bird(x) ∃y : tree1(y) ∧ live_ on(x,y) • ∃y tree1(y): ∀x bird(x) live_on(x,y) 88-680
FOPC – First Order Predicate Logic • Meaning Structure of Language • Choice of FOPC isn't completely arbitrary or driven by the needs of applications... • Human languages: • display a basic predicate-argument structure • make use of variables • make use of quantifers • display a partially compositional semantics 88-680
Predicate Argument Structure • בכל השפות המדוברות מבנה הפרדיקט-ארגומנטים הוא הגרעין והיסוד של המשמעות בשפה. • ההנחה הגסה הראשונה היא, אם כן, שבמשפט קיים יחס כלשהו בין המושגים המובעים במילים שבו וברכיביו השונים. • אותו יחס הוא הבסיס לייצוג הסמנטי שיש למשפט. • אחד התפקידים החשובים של התחביר הוא לתת מסגרת שתאפשר בנייה של אותו ייצוג, והייצוג צריך שיהיה לו קשר הדוק עם אותו מבנה תחבירי. 88-680
Predicate Argument Structure • Predicate-like elements: • Verbs, VPs, prepositions, adjectives, some nouns. • Argument-like elements: • Nouns, Nominals, NPs, etc 88-680
Example • “John gave Mary the book” • Giving(John, Mary, book) • ∃x,y: Giving(x)∧ Giver(John,x) ∧ Givee(Mary,x) ∧ Given(y,x) ∧ Book(y) 88-680
Semantic Analysis • Syntax driven semantics: • Most methods rely in some way on a prior or concurrent syntactic analysis (parse). • Compositional Semantics: • at the core of most methods is the principle of compositionality which states that the meaning of the whole is based on the meaning of parts. • What are the parts? Words and syntactic constructs 88-680
Augmented CF Rules • Verb serves{x y ∃e IsA(e, serving) ∧ server(e,y) ∧ served(e,x) } • S NP VP {VP.sem(NP.sem) } • VP Verb NP {Verb.sem(NP.sem)} • MassNoun Meat {Meat} • ProperNoun AyCaramba {AyCaramba} 88-680
Determiners 88-680
Issues • How to represent tense and aspect? • Idioms (are not compositional)! • Pragmatics (extra-syntactic)! • Inference • And much more… 88-680
Ambiguity Resolution • Still have problem of ambiguity resolution: • Word sense disambiguation • Attachment ambiguity resolution 88-680
WSD • Problem: many words have different meanings or senses, i.e., there is ambiguity about how they are to be specifically interpreted (e.g., differentiate). • Task: to determine which of the senses of an ambiguous word is invoked in a particular use of the word by looking at the context of its use. 88-680
Word Senses • Bank • The rising ground bordering a lake, river, or sea • An establishment for the custody, loan exchange, or issue of money, for the extension of credit, and for facilitating the transmission of funds • Title • Name/heading of a book, statue, work of art or music, etc. • Material at the start of a film • The right of legal ownership (of land) • The document that is evidence of the right • An appellation of respect attached to a person’s name • A written work (synecdoche: part stands for the whole) 88-680
Solution • look at the words around an ambiguous word in a large context window. Each word contributes potentially useful information about which sense of the ambiguous word is likely to be used with it. • Apply a supervised learning mechanism 88-680
WordNet • on-line lexical reference system, linguistically motivated • nouns, verbs, adjectives, adverbs are organized into synonym sets (lexical concepts), • contains a taxonomy of about 100,000 words 88-680
WordNet • semantic relationships represented • Synonymy: similarity of work meaning, concepts as sets of words • Antonymy: dichotomy in meaning of words, used for organizing adjectives and adverbs • Hyponymy:IS-A relationships Superconcept to Subconcepts • Meronymy: Part-Of relationships between concepts • Morphological Relationships: used to reduce word forms 88-680