Perceptually Grounded Selectional Preferences – Using Flickr Image and Video tags for Natural Language Semantics

Gerard de Melo
Gerard de MeloAssistant Professor at Rutgers University

Selectional preferences (SPs) are widely used in natural language processing as a rich source of semantic information. While SPs have been traditionally induced from textual data, human lexical acquisition is known to rely on both linguistic and perceptual experience. We present the first SP learning method that simultaneously draws knowledge from text, images and videos, using image and video descriptions to obtain visual features. Our results show that it outperforms linguistic and visual models in isolation, as well as the existing SP induction approaches.

Perceptually Grounded Selectional Preferences
Katia Shutova es407@cam.ac.uk
https://www.cl.cam.ac.uk/~es407/
Niket Tandon ntandon@mpi-inf.mpg.de
https://www.mpi-inf.mpg.de/~ntandon/
Gerard de Melo gdm@demelo.org
http://gerard.demelo.org
Contact
1. Philip Resnik (1993). Selection and information: A class-based approach to lexical relationships. Technical report, Univ. of Pennsylvania.
2. Frank Keller & Mirella Lapata (2003). Using the Web to obtain frequencies for unseen bigrams. Comp. Ling. 29(3):459–484.
3. Mats Rooth, Stefan Riezler, Detlef Prescher, Glenn Carroll, Franz Beil (1999). Inducing a semantically annotated lexicon via
EM-based clustering. Proc. ACL 1999.
4. Sebastian Pado, Ulrike Pado, Katrin Erk (2007). Flexible, corpus-based modelling of human plausibility judgements.
Proc. EMNLP-CoNLL 2007.
5. Diarmuid O ́Seaghdha (2010). Latent variable models of selectional preference. Proc. ACL 2010.
6. Ekaterina Shutova (2010) . Automatic metaphor interpretation as a paraphrasing task. Proc. NAACL 2010.
References
Selectional Preferences are semantic constraints of a predicate
on its arguments
The authors wrote a new paper. ✔ high plausibility
The paper wrote a new author. ✘ Very low plausibility
The cat is eating your sausage. ✔ high plausibility
The carrot is eating your keys. ✘ Very low plausibility
Knowledge of selectional preferences is useful in many NLP tasks:
●
Word Sense Disambiguation
●
Parsing (resolving attachments)
●
Semantic Role Labelling
●
Natural Language Inference
●
Detecting multi-word expressions
●
Etc.
What are Selectional Preferences?
Previous work uses purely text-based methods:
●
Problem of topic bias / figurative uses of words: E.g. “cut” mainly occurs with
“cost” and “price” as arguments in the BNC.
●
→ Skew towards abstract uses, different from our daily life experience of cutting
Our Approach: Use Multimodal Data
●
BNC for text (parsed using RASP parser)
●
100 million Flickr images/videos from Yahoo! Webscope Flickr-100M dataset
Challenge: From a set of Flickr Tags to noun–verb pairs
Collecting Multimodal Correlations
Step 1: Acquisition of Argument Classes
Observed data is sparse → Need to generalize
Spectral Clustering of nouns using Jensen-Shannon divergence as sim. measure
Step 2: Quantifying Selectional Preferences
Selectional Preference Model
Shutova (2010) approach: metaphor interpretation as paraphrasing
“a carelessly leaked report” → “a carelessly disclosed report”
1) Take maximum likelihood candidate verbs
2) Filter by semantic similarity to target verb
3) Filter for a strong selectional preference fit (assuming it indicates literalness or
conventionality) so as to remove figurative uses
Application to Metaphor Interpretation
Multimodal selectional preferences outperform
●
purely linguistic and visual models, and
●
previous state-of-the-art models
Conclusions
Method
Seen
Dataset
Unseen
Dataset
Rooth et al. (1999) EM 0.487 0.520
Pado et al. (2007)
VSM
0.490 0.430
O'Seaghda (2010) LDA 0.548 0.605
Visual Model 0.126 0.132
Linguistic Model 0.688 0.559
Interpolated Model 0.728 0.430
Direct Evaluation
mother
sitting
baby
lap
rachel lind
wristwatch
pajamas
Clothes
etc.
Ekaterina Shutova Niket Tandon Gerard de Melo
University of Cambridge Max Planck Institute
for Informatics
Tsinghua University
Shutova (2010) LSP ISP
Mean Avg. Prec. (MAP) on
Shutova (2010) gold data 0.62 0.62 0.65
Results on Keller & Lapata (2003)
Datasets (Spearman Rho)
Visual Features: verb lemmas
co-occurring with nouns
Linguistic Features:
grammatical relations
Approach
1) Stemming
2) Filtering:
Remove rare words
and named entities
3) POS tagging:
by jointly disambiguating
tags to WordNet synsets
so as to maximize
coherence
WordNet
priors
similarities
https://www.flickr.com/photos/seandreilinger/465827703/
canon
rebel
400D
ball
portfolio
yellow
serve
website
racket
roland
garros
etc.
https://www.flickr.com/photos/pysanchis/2521372121/

Recommended

SEMAC Graph Node Embeddings for Link Prediction by
SEMAC Graph Node Embeddings for Link PredictionSEMAC Graph Node Embeddings for Link Prediction
SEMAC Graph Node Embeddings for Link PredictionGerard de Melo
932 views39 slides
How to Manage your Research by
How to Manage your ResearchHow to Manage your Research
How to Manage your ResearchGerard de Melo
2.3K views142 slides
Knowlywood: Mining Activity Knowledge from Hollywood Narratives by
Knowlywood: Mining Activity Knowledge from Hollywood NarrativesKnowlywood: Mining Activity Knowledge from Hollywood Narratives
Knowlywood: Mining Activity Knowledge from Hollywood NarrativesGerard de Melo
848 views28 slides
Learning Multilingual Semantics from Big Data on the Web by
Learning Multilingual Semantics from Big Data on the WebLearning Multilingual Semantics from Big Data on the Web
Learning Multilingual Semantics from Big Data on the WebGerard de Melo
1.2K views156 slides
From Big Data to Valuable Knowledge by
From Big Data to Valuable KnowledgeFrom Big Data to Valuable Knowledge
From Big Data to Valuable KnowledgeGerard de Melo
1K views44 slides
Scalable Learning Technologies for Big Data Mining by
Scalable Learning Technologies for Big Data MiningScalable Learning Technologies for Big Data Mining
Scalable Learning Technologies for Big Data MiningGerard de Melo
1.7K views152 slides

More Related Content

More from Gerard de Melo

Information Extraction from Web-Scale N-Gram Data by
Information Extraction from Web-Scale N-Gram DataInformation Extraction from Web-Scale N-Gram Data
Information Extraction from Web-Scale N-Gram DataGerard de Melo
1.8K views75 slides
UWN: A Large Multilingual Lexical Knowledge Base by
UWN: A Large Multilingual Lexical Knowledge BaseUWN: A Large Multilingual Lexical Knowledge Base
UWN: A Large Multilingual Lexical Knowledge BaseGerard de Melo
1.1K views1 slide
Multilingual Text Classification using Ontologies by
Multilingual Text Classification using OntologiesMultilingual Text Classification using Ontologies
Multilingual Text Classification using OntologiesGerard de Melo
1.4K views18 slides
Extracting Sense-Disambiguated Example Sentences From Parallel Corpora by
Extracting Sense-Disambiguated Example Sentences From Parallel CorporaExtracting Sense-Disambiguated Example Sentences From Parallel Corpora
Extracting Sense-Disambiguated Example Sentences From Parallel CorporaGerard de Melo
1.7K views68 slides
Towards a Universal Wordnet by Learning from Combined Evidence by
Towards a Universal Wordnet by Learning from Combined EvidenceTowards a Universal Wordnet by Learning from Combined Evidence
Towards a Universal Wordnet by Learning from Combined EvidenceGerard de Melo
1.8K views65 slides
Not Quite the Same: Identity Constraints for the Web of Linked Data by
Not Quite the Same: Identity Constraints for the Web of Linked DataNot Quite the Same: Identity Constraints for the Web of Linked Data
Not Quite the Same: Identity Constraints for the Web of Linked DataGerard de Melo
989 views18 slides

More from Gerard de Melo(8)

Information Extraction from Web-Scale N-Gram Data by Gerard de Melo
Information Extraction from Web-Scale N-Gram DataInformation Extraction from Web-Scale N-Gram Data
Information Extraction from Web-Scale N-Gram Data
Gerard de Melo1.8K views
UWN: A Large Multilingual Lexical Knowledge Base by Gerard de Melo
UWN: A Large Multilingual Lexical Knowledge BaseUWN: A Large Multilingual Lexical Knowledge Base
UWN: A Large Multilingual Lexical Knowledge Base
Gerard de Melo1.1K views
Multilingual Text Classification using Ontologies by Gerard de Melo
Multilingual Text Classification using OntologiesMultilingual Text Classification using Ontologies
Multilingual Text Classification using Ontologies
Gerard de Melo1.4K views
Extracting Sense-Disambiguated Example Sentences From Parallel Corpora by Gerard de Melo
Extracting Sense-Disambiguated Example Sentences From Parallel CorporaExtracting Sense-Disambiguated Example Sentences From Parallel Corpora
Extracting Sense-Disambiguated Example Sentences From Parallel Corpora
Gerard de Melo1.7K views
Towards a Universal Wordnet by Learning from Combined Evidence by Gerard de Melo
Towards a Universal Wordnet by Learning from Combined EvidenceTowards a Universal Wordnet by Learning from Combined Evidence
Towards a Universal Wordnet by Learning from Combined Evidence
Gerard de Melo1.8K views
Not Quite the Same: Identity Constraints for the Web of Linked Data by Gerard de Melo
Not Quite the Same: Identity Constraints for the Web of Linked DataNot Quite the Same: Identity Constraints for the Web of Linked Data
Not Quite the Same: Identity Constraints for the Web of Linked Data
Gerard de Melo989 views
Good, Great, Excellent: Global Inference of Semantic Intensities by Gerard de Melo
Good, Great, Excellent: Global Inference of Semantic IntensitiesGood, Great, Excellent: Global Inference of Semantic Intensities
Good, Great, Excellent: Global Inference of Semantic Intensities
Gerard de Melo2K views
YAGO-SUMO: Integrating YAGO into the Suggested Upper Merged Ontology by Gerard de Melo
YAGO-SUMO: Integrating YAGO into the Suggested Upper Merged OntologyYAGO-SUMO: Integrating YAGO into the Suggested Upper Merged Ontology
YAGO-SUMO: Integrating YAGO into the Suggested Upper Merged Ontology
Gerard de Melo2.2K views

Recently uploaded

Initiating and Advancing Your Strategic GIS Governance Strategy by
Initiating and Advancing Your Strategic GIS Governance StrategyInitiating and Advancing Your Strategic GIS Governance Strategy
Initiating and Advancing Your Strategic GIS Governance StrategySafe Software
176 views68 slides
Elevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlue by
Elevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlueElevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlue
Elevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlueShapeBlue
222 views7 slides
LLMs in Production: Tooling, Process, and Team Structure by
LLMs in Production: Tooling, Process, and Team StructureLLMs in Production: Tooling, Process, and Team Structure
LLMs in Production: Tooling, Process, and Team StructureAggregage
42 views77 slides
State of the Union - Rohit Yadav - Apache CloudStack by
State of the Union - Rohit Yadav - Apache CloudStackState of the Union - Rohit Yadav - Apache CloudStack
State of the Union - Rohit Yadav - Apache CloudStackShapeBlue
297 views53 slides
Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And... by
Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And...Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And...
Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And...ShapeBlue
106 views12 slides
Updates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBIT by
Updates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBITUpdates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBIT
Updates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBITShapeBlue
206 views8 slides

Recently uploaded(20)

Initiating and Advancing Your Strategic GIS Governance Strategy by Safe Software
Initiating and Advancing Your Strategic GIS Governance StrategyInitiating and Advancing Your Strategic GIS Governance Strategy
Initiating and Advancing Your Strategic GIS Governance Strategy
Safe Software176 views
Elevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlue by ShapeBlue
Elevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlueElevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlue
Elevating Privacy and Security in CloudStack - Boris Stoyanov - ShapeBlue
ShapeBlue222 views
LLMs in Production: Tooling, Process, and Team Structure by Aggregage
LLMs in Production: Tooling, Process, and Team StructureLLMs in Production: Tooling, Process, and Team Structure
LLMs in Production: Tooling, Process, and Team Structure
Aggregage42 views
State of the Union - Rohit Yadav - Apache CloudStack by ShapeBlue
State of the Union - Rohit Yadav - Apache CloudStackState of the Union - Rohit Yadav - Apache CloudStack
State of the Union - Rohit Yadav - Apache CloudStack
ShapeBlue297 views
Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And... by ShapeBlue
Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And...Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And...
Enabling DPU Hardware Accelerators in XCP-ng Cloud Platform Environment - And...
ShapeBlue106 views
Updates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBIT by ShapeBlue
Updates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBITUpdates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBIT
Updates on the LINSTOR Driver for CloudStack - Rene Peinthor - LINBIT
ShapeBlue206 views
Setting Up Your First CloudStack Environment with Beginners Challenges - MD R... by ShapeBlue
Setting Up Your First CloudStack Environment with Beginners Challenges - MD R...Setting Up Your First CloudStack Environment with Beginners Challenges - MD R...
Setting Up Your First CloudStack Environment with Beginners Challenges - MD R...
ShapeBlue173 views
DRaaS using Snapshot copy and destination selection (DRaaS) - Alexandre Matti... by ShapeBlue
DRaaS using Snapshot copy and destination selection (DRaaS) - Alexandre Matti...DRaaS using Snapshot copy and destination selection (DRaaS) - Alexandre Matti...
DRaaS using Snapshot copy and destination selection (DRaaS) - Alexandre Matti...
ShapeBlue139 views
"Package management in monorepos", Zoltan Kochan by Fwdays
"Package management in monorepos", Zoltan Kochan"Package management in monorepos", Zoltan Kochan
"Package management in monorepos", Zoltan Kochan
Fwdays33 views
Mitigating Common CloudStack Instance Deployment Failures - Jithin Raju - Sha... by ShapeBlue
Mitigating Common CloudStack Instance Deployment Failures - Jithin Raju - Sha...Mitigating Common CloudStack Instance Deployment Failures - Jithin Raju - Sha...
Mitigating Common CloudStack Instance Deployment Failures - Jithin Raju - Sha...
ShapeBlue180 views
CloudStack Managed User Data and Demo - Harikrishna Patnala - ShapeBlue by ShapeBlue
CloudStack Managed User Data and Demo - Harikrishna Patnala - ShapeBlueCloudStack Managed User Data and Demo - Harikrishna Patnala - ShapeBlue
CloudStack Managed User Data and Demo - Harikrishna Patnala - ShapeBlue
ShapeBlue135 views
"Running students' code in isolation. The hard way", Yurii Holiuk by Fwdays
"Running students' code in isolation. The hard way", Yurii Holiuk "Running students' code in isolation. The hard way", Yurii Holiuk
"Running students' code in isolation. The hard way", Yurii Holiuk
Fwdays36 views
The Power of Heat Decarbonisation Plans in the Built Environment by IES VE
The Power of Heat Decarbonisation Plans in the Built EnvironmentThe Power of Heat Decarbonisation Plans in the Built Environment
The Power of Heat Decarbonisation Plans in the Built Environment
IES VE79 views
"Surviving highload with Node.js", Andrii Shumada by Fwdays
"Surviving highload with Node.js", Andrii Shumada "Surviving highload with Node.js", Andrii Shumada
"Surviving highload with Node.js", Andrii Shumada
Fwdays56 views
Zero to Cloud Hero: Crafting a Private Cloud from Scratch with XCP-ng, Xen Or... by ShapeBlue
Zero to Cloud Hero: Crafting a Private Cloud from Scratch with XCP-ng, Xen Or...Zero to Cloud Hero: Crafting a Private Cloud from Scratch with XCP-ng, Xen Or...
Zero to Cloud Hero: Crafting a Private Cloud from Scratch with XCP-ng, Xen Or...
ShapeBlue198 views
NTGapps NTG LowCode Platform by Mustafa Kuğu
NTGapps NTG LowCode Platform NTGapps NTG LowCode Platform
NTGapps NTG LowCode Platform
Mustafa Kuğu423 views
Business Analyst Series 2023 - Week 4 Session 8 by DianaGray10
Business Analyst Series 2023 -  Week 4 Session 8Business Analyst Series 2023 -  Week 4 Session 8
Business Analyst Series 2023 - Week 4 Session 8
DianaGray10123 views
The Power of Generative AI in Accelerating No Code Adoption.pdf by Saeed Al Dhaheri
The Power of Generative AI in Accelerating No Code Adoption.pdfThe Power of Generative AI in Accelerating No Code Adoption.pdf
The Power of Generative AI in Accelerating No Code Adoption.pdf
Saeed Al Dhaheri32 views

Perceptually Grounded Selectional Preferences – Using Flickr Image and Video tags for Natural Language Semantics

  • 1. Perceptually Grounded Selectional Preferences Katia Shutova es407@cam.ac.uk https://www.cl.cam.ac.uk/~es407/ Niket Tandon ntandon@mpi-inf.mpg.de https://www.mpi-inf.mpg.de/~ntandon/ Gerard de Melo gdm@demelo.org http://gerard.demelo.org Contact 1. Philip Resnik (1993). Selection and information: A class-based approach to lexical relationships. Technical report, Univ. of Pennsylvania. 2. Frank Keller & Mirella Lapata (2003). Using the Web to obtain frequencies for unseen bigrams. Comp. Ling. 29(3):459–484. 3. Mats Rooth, Stefan Riezler, Detlef Prescher, Glenn Carroll, Franz Beil (1999). Inducing a semantically annotated lexicon via EM-based clustering. Proc. ACL 1999. 4. Sebastian Pado, Ulrike Pado, Katrin Erk (2007). Flexible, corpus-based modelling of human plausibility judgements. Proc. EMNLP-CoNLL 2007. 5. Diarmuid O ́Seaghdha (2010). Latent variable models of selectional preference. Proc. ACL 2010. 6. Ekaterina Shutova (2010) . Automatic metaphor interpretation as a paraphrasing task. Proc. NAACL 2010. References Selectional Preferences are semantic constraints of a predicate on its arguments The authors wrote a new paper. ✔ high plausibility The paper wrote a new author. ✘ Very low plausibility The cat is eating your sausage. ✔ high plausibility The carrot is eating your keys. ✘ Very low plausibility Knowledge of selectional preferences is useful in many NLP tasks: ● Word Sense Disambiguation ● Parsing (resolving attachments) ● Semantic Role Labelling ● Natural Language Inference ● Detecting multi-word expressions ● Etc. What are Selectional Preferences? Previous work uses purely text-based methods: ● Problem of topic bias / figurative uses of words: E.g. “cut” mainly occurs with “cost” and “price” as arguments in the BNC. ● → Skew towards abstract uses, different from our daily life experience of cutting Our Approach: Use Multimodal Data ● BNC for text (parsed using RASP parser) ● 100 million Flickr images/videos from Yahoo! Webscope Flickr-100M dataset Challenge: From a set of Flickr Tags to noun–verb pairs Collecting Multimodal Correlations Step 1: Acquisition of Argument Classes Observed data is sparse → Need to generalize Spectral Clustering of nouns using Jensen-Shannon divergence as sim. measure Step 2: Quantifying Selectional Preferences Selectional Preference Model Shutova (2010) approach: metaphor interpretation as paraphrasing “a carelessly leaked report” → “a carelessly disclosed report” 1) Take maximum likelihood candidate verbs 2) Filter by semantic similarity to target verb 3) Filter for a strong selectional preference fit (assuming it indicates literalness or conventionality) so as to remove figurative uses Application to Metaphor Interpretation Multimodal selectional preferences outperform ● purely linguistic and visual models, and ● previous state-of-the-art models Conclusions Method Seen Dataset Unseen Dataset Rooth et al. (1999) EM 0.487 0.520 Pado et al. (2007) VSM 0.490 0.430 O'Seaghda (2010) LDA 0.548 0.605 Visual Model 0.126 0.132 Linguistic Model 0.688 0.559 Interpolated Model 0.728 0.430 Direct Evaluation mother sitting baby lap rachel lind wristwatch pajamas Clothes etc. Ekaterina Shutova Niket Tandon Gerard de Melo University of Cambridge Max Planck Institute for Informatics Tsinghua University Shutova (2010) LSP ISP Mean Avg. Prec. (MAP) on Shutova (2010) gold data 0.62 0.62 0.65 Results on Keller & Lapata (2003) Datasets (Spearman Rho) Visual Features: verb lemmas co-occurring with nouns Linguistic Features: grammatical relations Approach 1) Stemming 2) Filtering: Remove rare words and named entities 3) POS tagging: by jointly disambiguating tags to WordNet synsets so as to maximize coherence WordNet priors similarities https://www.flickr.com/photos/seandreilinger/465827703/ canon rebel 400D ball portfolio yellow serve website racket roland garros etc. https://www.flickr.com/photos/pysanchis/2521372121/