Download presentation
Presentation is loading. Please wait.
Published byMitchell Harvey Modified over 9 years ago
1
RANLP, Borovets 27-29 Sept. 2007 Evaluating Algorithms for GRE (Going beyond Toy Domains) Ielka van der Sluis Albert Gatt Kees van Deemter University of Aberdeen, Scotland, UK
2
RANLP, Borovets 27-29 Sept. 2007 Outline GRE: Generation of Referring Expressions TUNA project: Corpus and Annotation Evaluation of Algorithms –Furniture Domain –People Domain [ Evaluation in the real world: STEC ]
3
RANLP, Borovets 27-29 Sept. 2007 TUNA project (ended Feb. 2007) TUNA: Towards a UNified Algorithm for Generating Referring Expressions. 1.Extend coverage of GRE algorithms (plurals, negation, gradable properties,…) 2.Improve empirical foundations of GRE Focus on –Content Determination –“First mention” NPs (no anaphora!)
4
RANLP, Borovets 27-29 Sept. 2007 TUNA results Elsewhere: –Reference to sets (e.g., Gatt 2006, 2007) –Gradable/vague properties (Van Deemter 2006) –Pointing (Van der Sluis & Krahmer 2007) –Large domains (Paraboni et al. 2007) This talk: empirical issues –Testing classic algorithms –Method: compute similarity to human-generated NPs
5
RANLP, Borovets 27-29 Sept. 2007 Method (overview) Elicitation experiment Leads to transparent corpus of referring expressions: –referent and distractors are known –Domain attributes are known Transparent corpora can be used for many purposes This talk: Compare some classic algorithms –giving each algorithm the same input as subjects –computing how similar algorithm’s output is to subjects’ output –We count semantic content only
6
RANLP, Borovets 27-29 Sept. 2007 Elicitation Experiment Furniture (simple domain) –TYPE, COLOUR, SIZE, ORIENTATION People (complex domain) –Nine annotated properties in total Location: –Vertical location (Y-DIMENSION) –Horizontal location (X-DIMENSION) the green desk facing backwards the sofa and the desk which are red the young man with a white shirt the man with the funny haircut the man on the left the chair in the top right
7
RANLP, Borovets 27-29 Sept. 2007 Furniture trial
8
RANLP, Borovets 27-29 Sept. 2007 People trial
9
RANLP, Borovets 27-29 Sept. 2007 Corpus setup Each corpus was carefully balanced, e.g. between singulars and plurals. Between-subjects design: -Location: Subjects discouraged from using locative expressions. +Location: Subjects not discouraged. -FaultCritical: Subjects could correct their utterances +FaultCritical: Subjects could not correct their utterances After discounting outliers and (self-reported) non-fluent speakers, 45 subjects were left
10
RANLP, Borovets 27-29 Sept. 2007 Experiment design: Furniture (-Location) 18 trials: (C=Colour, O=orientation, S-size) –1 referent: minimal identification uses {c}, {o}, {s}, {c,o}, {c,s}, or {o,s} [6 trials] –2 “similar” referents {c}, {o}, {s}, {c,o}, {c,s}, or {o,s} [6 trials] –2 “dissimilar referents” {c}, {o}, {s}, {c,o}, {c,s}, or {o,s} [6 trials]
11
RANLP, Borovets 27-29 Sept. 2007 Classic GRE Algorithms Full Brevity (FB; Dale 1989) –Generation of a minimal description Greedy Algorithm (GR; Dale 1989) –Always add property that removes the most distractors Incremental Algorithm (IA; Dale and Reiter 1995) –Add next useful property from an ordered list of properties. (“Preference Order” = PO)
12
RANLP, Borovets 27-29 Sept. 2007 Other evaluation studies Jordan 2000, Jordan & Walker 2005 –More than just identification (Jordan 2000) Siddharthan & Copestake 2004 –References in linguistic context Gupta & Stent 2005 –Realisation mixed with Content Determination Viethen & Dale 2006 –Only Colour and Location
13
RANLP, Borovets 27-29 Sept. 2007 Other evaluation studies General limitations: Limited numbers of subjects/referents Few attempts at balancing the corpus. (E.g., Viethen & Dale 2006 let subjects decide what to refer to.) IA: no teasing apart of preference orders
14
RANLP, Borovets 27-29 Sept. 2007 Extensions to the classics Plurality: (van Deemter 2002) –Extend each algorithm to search through disjunctions of increasing length Location: (van Deemter 2006) –Locatives treated as gradable: “the leftmost table/person” –E.g., suppose the referent x is located in column 3 => “x is left of column 4”, “x is left of column 5” … => “x is right of column 2”, “x is right of column 1”… Type: –People tend to use TYPE (Dale & Reiter 1995) –Here: All algorithms added TYPE.
15
RANLP, Borovets 27-29 Sept. 2007 Evaluation aims Hypothesis in Dale & Reiter 1995: –IA resembles human output most Our main questions: –Is this true? –How important are parameters (PO) for the IA? More generally: –assess ‘quality’ of classic GRE algorithms : –calculate average match between the description generated by an algorithm and the descriptions produced by people (for the same referent)
16
RANLP, Borovets 27-29 Sept. 2007 Evaluation metric Dice Coefficient: 2 x |Common properties| |total properties| A coefficient result of 1 indicates identical sets. 0 means no common terms We also used this to measure agreement between annotators of the corpus
17
RANLP, Borovets 27-29 Sept. 2007 (Assumptions behind DICE) Deletion of a property is slightly worse than addition of a property The discriminatory power of a description does not matter All properties are equidistant See Gatt & Van Deemter 2007, “Content Determination in GRE: evaluating the evaluator” )
18
RANLP, Borovets 27-29 Sept. 2007 Evaluation (I): Furniture Which preference orders for the IA? –Psycholinguistic evidence: COLOUR >> {ORIENTATION, SIZE} (Pechmann 89; Eikmeyer & Ahlsen 96; Belke & Meyer 02) Y-DIMENSION >> X-DIMENSION (Bryant et al, 1992; Arts 2004) Split data: +LOCATION vs –LOCATION This talk: focus on –LOCATION –LOCATION = approx. 800 descriptions Compare algorithms to a randomized IA (RAND)
19
RANLP, Borovets 27-29 Sept. 2007 Furniture: -LOCATION Significant FB/GR
20
RANLP, Borovets 27-29 Sept. 2007 Beyond Toy Domains More on Furniture corpus: Gatt et al. (ENLG-2007) With complex real-world objects: –Many different attributes can be used –Number of PO’s explodes –Few psycholinguistic precedents People domain attributes: –{ hasBeard, hasGlasses, age, hasTie, hasSuit, hasSuit, hasHair, hairColour, orientation } –9 Attributes, so 9! = 362880 possible POs
21
RANLP, Borovets 27-29 Sept. 2007 IA: Preference Orders for People Domain Little psycholinguistic evidence for choosing between all 362880 possible PO’s Focus on the most frequent Attributes: G=hasGlasses, B=hasBeard, H=hasHair, C=haircolour –Assumption: H and B must precede C –This leaves us with eight POs: { GBHC, GHBC,HBGC,HBCG, HGBC,BHGC, BHCG, BGHC }
22
RANLP, Borovets 27-29 Sept. 2007 Preference Orders and frequency Mean (std)Sum type1.39475 hasGlasses.68231 hasBeard.66226 HairColour.61210 hasHair.46158 orientation.2173 age.1034 hasTie.0412 hasSuit.014 hasShirt.013 For attributes other than {G,C,H,B}, we let corpus frequency determine the order E.g, IA-GBHC uses type, G,B,H,C, age, hasTie, hasSuit,hasShirt as its PO
23
RANLP, Borovets 27-29 Sept. 2007 Results People Domain IA-BASE Significant Significant by subjects GR
24
RANLP, Borovets 27-29 Sept. 2007 Results People domain IA_base performs very badly now So much about the best IA’s that start with {B,H,G,C} and end with Some of these did much worse: –IA_BHCG had DICE=0.6, making it significantly worse (by subjects) than GR!
25
RANLP, Borovets 27-29 Sept. 2007 Summary People domain gives much lower DICE scores than Furniture domain Difference between “good” and “bad” POs was enormous in People domain
26
RANLP, Borovets 27-29 Sept. 2007 Summary The “Incremental Algorithm” (IA): –not an algorithm but a class of algorithms The best IA beats all other algorithms, but the worst is very bad... GR performs remarkably well. How to choose a suitable PO? –Furniture: few attributes; psycholinguistic precedent Still, there is variation. –People: more attributes; no precedents Variation even greater!
27
RANLP, Borovets 27-29 Sept. 2007 Discussion Suppose you want to build a GRE algorithm for a new and complex domain, for which no transparent corpus is available. Psycholinguistic principles are unlikely to help you much If corpus is also not balanced, then frequency doesn’t say much either …
28
RANLP, Borovets 27-29 Sept. 2007 Other uses of this method: STEC Summer 2007: First NLG Shared task Evaluation Challenge (STEC) STEC involved GRE only, focussing on Content Determination 22 GRE Algorithms were submitted and evaluated (6 teams) Reported in UCNLG+MT workshop, Copenhagen, Sept 2007
29
RANLP, Borovets 27-29 Sept. 2007 Other uses of this corpus: STEC Each algorithm was compared with the TUNA corpus (minus 40% training set) –Both Furniture and People domain –DICE measured “humanlikeness” –Singulars only Each algorithm was also tested in terms of identification time (by human reader)
30
RANLP, Borovets 27-29 Sept. 2007 Other uses of this corpus: STEC Future STEC: –beyond “first mention” –beyond Content Determination –more hearer-oriented experiments
31
RANLP, Borovets 27-29 Sept. 2007 STEC results 1.The more minimal the descriptions generated by these 22 systems were, the worse their DICE scores were
32
RANLP, Borovets 27-29 Sept. 2007
33
RANLP, Borovets 27-29 Sept. 2007 2. No relation between humanlikeness and identification time –Best system in terms of DICE was worst- but-one in terms of identification time More research needed on the different criteria for judging NLG output
34
RANLP, Borovets 27-29 Sept. 2007 Thank you
35
RANLP, Borovets 27-29 Sept. 2007 Annotator agreement Semantic markup was applied manually to all descriptions in the corpus. 2 annotators were given a stratified random sample Comparison used Dice. meanmode Furniture0.89 (A/B) 1 (71.1%) Annotator A0.93 (A/us) 1 (74.4%) Annotator B0.92 (B/us) 1(73%) People0.89 (A/B) 1(70%) Annotator A0.84 (A/us) 1(41.1%) Annotator B.78 (B/us) 1(36.3%)
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.