25
Discussion Slides for Computational Semantics Reading Group 13 January 2010 Dustin Smith [email protected] Deriving Narrative Morphologies via Analogical Story Merging . Mark Alan Finlayson. 2009 Thursday, January 14, 2010

Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Discussion Slides for Computational Semantics Reading Group

13 January 2010

Dustin [email protected]

Deriving Narrative Morphologies via Analogical Story Merging. Mark Alan

Finlayson. 2009

Thursday, January 14, 2010

Page 2: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

- What are they? - stories, folk-tales, fairy tales, fables, myths, - memes: transmitted socially and evolve

- What purpose? - Medium for transmitting cultural values, norms, and beliefs - For remembering coherent structures of stories [Frederic Bartlett; Roger Schank]

Premise: overarching narrative semantics may follow a template. Discourse level, or above discourse?

Narratives

Thursday, January 14, 2010

Page 3: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Thursday, January 14, 2010

Page 4: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Narrative Morphology: the set of plot elements and “valid” combinations [Vladamir Propp, 1928] - acquired through analyzing the belief/knowledge patterns conveyed in a cultureʼs narratives. - can be formalized as a grammar

Across cultures, narrative morphologies share top structure: - preparation, motivation, main actions, conclusion - narrative elements are classified by their “function” [?]

Question: can these automatically be derived from an annotated corpus?

Narrative Morphology

Thursday, January 14, 2010

Page 5: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Input: Semantically annotated stories (5 Shakespearean stories)

Output: Finite state machine model

- Using Bayesian Model Merging

1. Construct Initial Morphology, M0: - represent each story as a sequence of propositional states2. Define the state-merge operation - replaces two states with a single state - state emissions are observations, probabilities evenly divided3. Define prior P(M) 4. Search through space of FSMs, merging similar states

Analogical Story Merging

Thursday, January 14, 2010

Page 6: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

story posteriorBayesian Model Merging

[Andreas Stolcke and Steven Omohundro 1994]Thursday, January 14, 2010

Page 7: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

story posteriorBayesian Model Merging

[Andreas Stolcke and Steven Omohundro 1994]

Merge state nodes to

generate Mi+1

Thursday, January 14, 2010

Page 8: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

story posteriorBayesian Model Merging

select model with max posterior!

[Andreas Stolcke and Steven Omohundro 1994]

Merge state nodes to

generate Mi+1

Thursday, January 14, 2010

Page 9: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

[Patrick Winston 1981 ]

Example Input Stories

- Simplified summary of Macbeth:

- Summaries contain between 7-11 events.- These are converted to “morphological annotation” [?]: e.g., sister_predicts_murder(MAcbeth)

Thursday, January 14, 2010

Page 10: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Representation of States

Thursday, January 14, 2010

Page 11: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Thursday, January 14, 2010

Page 12: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Thursday, January 14, 2010

Page 13: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Thursday, January 14, 2010

Page 14: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Defining the modelsʼ priors

Structure Mapping Engine

- Parameter p = 0.95, controls delicate balance between more compact models and modelʼs likelihood to generate data.

- Each state has multiple morphological events,probability of each state is 1 if all events in the state are pairwise similar and 0 otherwise (omitted events kills it?)

- suggestions [p 135]: account for causal/temporal relations in similarity function. Generalization of predicates or terms?

Thursday, January 14, 2010

Page 15: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Model learnt from Macbeth, Hamlet, Julius Caesar, Othello and Taming of the Shrew

Thursday, January 14, 2010

Page 16: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Model learnt from Macbeth, Hamlet, Julius Caesar, Othello and Taming of the Shrew

{2,3,6}: other character persuades protagonist to murder enemy

Thursday, January 14, 2010

Page 17: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

17

Some discussion questionsAbout the research objective: - What are the next steps for the inferred structures? - Are morphologies/Plot Units, cognitively plausible?

About the model: - The problem of granularity: sequences are not flat, but have different levels of is-a/part-of abstraction. - The problem of missing information?

About the name “narrative” versus “text”, “story”, etc. - Does this kind of inference apply to blogs, openmind, - Controlled-English Shakespeare vs. commonsense scripts?

Thursday, January 14, 2010

Page 18: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

18

Story Understanding?

Figure 1. Story understanding software, enjoying itself.Thursday, January 14, 2010

Page 19: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

19

OpenMind Indoor Commonsense

- Honda’s OMICS project [1], a clone of the Media Lab effort. Collecting knowledge exclusively about indoor common sense. High quality, manually reviewed.

- Has a parallel corpus of stories about ways to accomplish 819 different goals.

- OMICS database currently has 1,172,971 items from 3K users.

[1] http://openmind.hri-us.com/

Thursday, January 14, 2010

Page 20: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

20

access the internetact as a security guardanswer the doorbellanswer the phoneapply band aidassist person standing upassist someone in walkingboil the milkbuy from vending machinecall 911calm an infantchange a baby diaperchange a bulbchange bed sheetscharge a cell phonecheck for intruderscheck for weathercheck if a store is openchop vegetablesclean a spillclean the dishesclean the floorclean the showerclean the tableclean upclean up toysclose the blindsclose the curtainscook fishcook noodlecook pastacook ricedance with the childrendo laundrydraw the curtains

dry clothesdust an objectempty the kitchen sinkempty the trashentertain childrenerase the whiteboardfeed a childfeed a pet catfeed a pet dogfeed infantfeed the fishfetch a cold drinkfetch a ladderfetch an objectfill water in containerfind a personfind an objectfind out more informationfind the timefold clothesfollow someone aroundgather all scattered toysget food from refrigeratorget mailget the newspapergive a medicinegive a messagegive a message on phonego outsidegreet a visitorguard the househandle toxic materialshang clothesheat food in microwaveheat food on kitchen gas

help someone carry thingiron clotheskeep the dog awaykick a ballload the dishwasherlock up the houselock windowsmail a lettermake a bedmake a dinner reservationmake a flight reservationmake a listmake a presentationmake a shopping listmake a tossed saladmake baby sleepmake breakfastmake coffeemake fresh orange juicemake hot dogmake soupmake sure children fedmake teamake toasted breadmaking omelettemove furnituremow the lawnopen a web pageopen packageopen the garageopen the mailpack a mailing boxpack a suitcasepaint a wallpay bills

perform research on specphotocopy a paperpick up dishesplace ladder near wallplay a game on the compplay a movieplay a songplay pianoplug an electric appliance in plug battery into chargerpour beer into a glassprint documentpush someone in a wheelpush somethingput away groceriesput object awayput up a paintingraise the blindsread a story to a childrecharge batteriesremove and replace garbagereplace a refrigerator filterreplace a water tap filterreplace batteries in the treplace heater filterretrieve a toolsecure all exitssecure all windowssecure the perimeter of send a faxsend party invitationsserve a drinkserve a mealset a wake up alarmset the dining table

Thursday, January 14, 2010

Page 21: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

21

Four example parsed narratives (of 36)

Thursday, January 14, 2010

Page 22: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

22

... and related background knowledge

Thursday, January 14, 2010

Page 23: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Relational Markov Models

C. Anderson, P. Domingos, and D.S. Weld. Relational Markov Models and their Application to Adaptive Web Naviation. 2002.

A graph rank-based heuristic

on the state transitions is used

to generalize arguments,

in favor of a compact representation.

Thursday, January 14, 2010

Page 24: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Relational Sequence Alignment

[1] Much related work by Kristian Kersting and Luc De Raedt.

- Relational (in first order logic), instead of propositional, descriptions of states.

Predicates/arity: vi/2, cd/1, ls/0, pdfview/2

Ground atoms -- predicates with non-variable terms:

vi(ch2,tex)

Ground clauses:

Generalized Clauses (with variables):

cd(X), vi(Y, tex), latex(Y, tex)

Thursday, January 14, 2010

Page 25: Deriving Narrative Morphologies via Analogical Story ...alumni.media.mit.edu/~dustin/mr/presentations/dsmith-20100103.pdf · 3/1/2010  · Narrative Morphology: the set of plot elements

Sequential

Relational Sequence Alignment

[1] Much related work by Kristian Kersting and Luc De Raedt.Thursday, January 14, 2010