SlideShare a Scribd company logo
1 of 23
Download to read offline
Deep Semantic Learning
for Conversational Agents
Candidate: Martino Mensio
Supervisor: Maurizio Morisio
Tutor: Giuseppe Rizzo
12 April 2018
1
Objectives
1. Identify the approaches to build a
Conversational Agent with Natural
Language Understanding
2. Use the context of interaction
2
Background
3
Background: Conversational Agents
What they can do:
- automated interaction with customer
- virtual assistants
What content they can provide:
- Chit-chat (small talk)
- Goal-oriented
- Knowledge-based
4
Background: from questions to answers
5
Background: an example of Understanding
6
Background: Recurrent Neural Networks
7
Background: intent classification
[1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot
filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association.
8
Background: slot filling
[1] Liu, B. and Lane, I. (2016). Attention-based recurrent
neural network models for joint intent detection and slot
filling. Proceedings of The 17th Annual Meeting of the
International Speech Communication Association.
9
Background: Word Embeddings
[2] Harris, Z. S. (1970). Distributional structure. In Papers in structural and transformational linguistics (pp.
775-794). Springer, Dordrecht.
10
Distributional Semantics [2]: words used in similar
contexts have similar meaning
- each word corresponds to a vector of reals
- small dimensionality (50~300)
- semantic distribution in a multidimensional space
The approach
11
Approach: the multi-turn interactions
- detect the change of intent
- capture intent dependencies
- consider the agent words 12
Approach: difference between multi-turn and single-turn
13
Approach: multi-turn example
14
Approach: Word Embeddings for Italian language
recomputation of Italian Wikipedia embeddings
with proper tokenization (with respect to [6])
15[6] Berardi, G., Esuli, A., & Marcheggiani, D. (2015). Word Embeddings Go to Italy: A Comparison of Models and
Training Datasets. In IIR.
“Voglio una bici vicino a piazza castello, grazie”
↓
[“Voglio”, “una”, “bici”, “vicino”, “a”, “piazza”, “castello”, “,”, “grazie”]
Results
16
Results: the datasets
available:
- ATIS (single-turn) [3]
- nlu-benchmark (single-turn) [4]
- kvret (multi-turn) [5]
collected:
- bikes Italian (single-turn)
- bikes English (single-turn)
17
[3] Hemphill, C., Godfrey, J., Doddington, G. (1990). The ATIS spoken language systems pilot corpus. DARPA Speech
and Natural Language Workshop
[4] https://github.com/snipsco/nlu-benchmark
[5] Eric, M. and Manning, C. (2017). Key-value retrieval networks for task-oriented dialogue. SIGDIAL 2017: Session
on Natural Language Generation for Dialog Systems
Results: multi-turn intent classification
results on kvret dataset [5]
18
approach
F1 epoch number
intent RNN agent words
✓ LSTM ✓ 0.9987 7
✓ LSTM ✘ 0.9987 8
✓ GRU ✓ 0.9975 14
✘ ✓ 0.9951 5
✓ GRU ✘ 0.9585 9
[1]✘ ✘ 0.8524 8
[1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot
filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association.
[5] Eric, M. and Manning, C. (2017). Key-value retrieval networks for task-oriented dialogue. SIGDIAL 2017: Session
on Natural Language Generation for Dialog Systems
Results: Italian Word Embeddings
19
[7] Mikolov, T., Yih, W. T., & Zweig, G. (2013). Linguistic regularities in continuous space word representations. In
Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics:
Human Language Technologies (pp. 746-751).
Word Embeddings accuracy
Italian values from [6] on Wikipedia 44.81%
Computed Italian values on Wikipedia 58.14%
analogy test [7]:
- semantic (capital-country, nationality adjective, currency, family)
- syntactic (m-f, singular-plural, tenses, comparatives, superlatives)
[6] Berardi, G., Esuli, A., & Marcheggiani, D. (2015). Word Embeddings Go to Italy: A Comparison of Models and
Training Datasets. In IIR.
Results: the difference on the global tasks (Italian)
Measured on the bike sharing dataset on the
approach by [1]
20
Word Embeddings intent classification F1 slot filling F1
Italian values from [6] on Wikipedia,
730k vectors
0.8421 0.5666
Computed Italian values on Wikipedia,
758k vectors
0.8947 0.6153
[7] Berardi, G., Esuli, A., & Marcheggiani, D. (2015). Word Embeddings Go to Italy: A Comparison of Models and
Training Datasets. In IIR.
[1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot
filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association.
Results: the difference of embeddings on the two tasks (English)
21
Embeddings
intent classification F1 slot filling F1
ATIS nlu-bench
mark
bikes
english
ATIS nlu-bench
mark
bikes
english
Trainable, random
initialization
0.9740 0.9928 0.9428 0.9425 0.9177 0.9000
[8] precomputed,
685k keys,
20k unique vectors
0.9660 0.9928 0.9714 0.9588 0.8970 0.9375
[8] precomputed,
685k keys,
685k unique vectors
0.9860 0.9928 0.9714 0.9649 0.9170 0.9689
[1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot
filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association.
[8] https://spacy.io/models/en
Measured on the approach by [1]
Conclusions
- results of the multi-turn show the
importance of context
- results for the word embeddings show the
importance of their proper choice
22
Future works
- multi-turn slot filling to remove
handcrafted dialog tracking
23

More Related Content

Similar to Deep Semantic Learning for Conversational Agents

Toward FAIR Semantic Resources
Toward FAIR Semantic ResourcesToward FAIR Semantic Resources
Toward FAIR Semantic ResourcesEUDAT
 
A Presentation of My Research Activity
A Presentation of My Research ActivityA Presentation of My Research Activity
A Presentation of My Research ActivityRoberto Casadei
 
Collaboration and improvisation
Collaboration and improvisationCollaboration and improvisation
Collaboration and improvisationChristian Voigt
 
DFlow is dead. Long live Tako!
DFlow is dead. Long live Tako!DFlow is dead. Long live Tako!
DFlow is dead. Long live Tako!Roberto Minelli
 
Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013 Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013 Azahara TIC
 
Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013The Pink Salmon
 
To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...
To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...
To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...Pablo Aragón
 
SATANJEEV BANERJEE
SATANJEEV BANERJEESATANJEEV BANERJEE
SATANJEEV BANERJEEbutest
 
APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...
APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...
APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...AishwaryaChemate
 
A Survey on Sentiment Analysis and Opinion Mining.pdf
A Survey on Sentiment Analysis and Opinion Mining.pdfA Survey on Sentiment Analysis and Opinion Mining.pdf
A Survey on Sentiment Analysis and Opinion Mining.pdfMandy Brown
 
A Survey On Sentiment Analysis And Opinion Mining Techniques
A Survey On Sentiment Analysis And Opinion Mining TechniquesA Survey On Sentiment Analysis And Opinion Mining Techniques
A Survey On Sentiment Analysis And Opinion Mining TechniquesSabrina Green
 
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...ijnlc
 
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...kevig
 
[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...
[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...
[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...University of Bologna
 
Discussing about information architecture for language learning: apps and str...
Discussing about information architecture for language learning: apps and str...Discussing about information architecture for language learning: apps and str...
Discussing about information architecture for language learning: apps and str...Flavio Nazario
 
Crowdsourcing Documentation in Software Engineering
Crowdsourcing Documentation in Software EngineeringCrowdsourcing Documentation in Software Engineering
Crowdsourcing Documentation in Software EngineeringMargaret-Anne Storey
 
New Research Articles 2020 May Issue International Journal of Software Engin...
New Research Articles 2020 May  Issue International Journal of Software Engin...New Research Articles 2020 May  Issue International Journal of Software Engin...
New Research Articles 2020 May Issue International Journal of Software Engin...ijseajournal
 

Similar to Deep Semantic Learning for Conversational Agents (20)

Toward FAIR Semantic Resources
Toward FAIR Semantic ResourcesToward FAIR Semantic Resources
Toward FAIR Semantic Resources
 
First draft
First draftFirst draft
First draft
 
A Presentation of My Research Activity
A Presentation of My Research ActivityA Presentation of My Research Activity
A Presentation of My Research Activity
 
Collaboration and improvisation
Collaboration and improvisationCollaboration and improvisation
Collaboration and improvisation
 
DFlow is dead. Long live Tako!
DFlow is dead. Long live Tako!DFlow is dead. Long live Tako!
DFlow is dead. Long live Tako!
 
Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013 Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013
 
Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013Comunicación presentada en Disconference 2013
Comunicación presentada en Disconference 2013
 
To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...
To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...
To Thread or Not to Thread: The Impact of Conversation Threading on Online Di...
 
SATANJEEV BANERJEE
SATANJEEV BANERJEESATANJEEV BANERJEE
SATANJEEV BANERJEE
 
APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...
APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...
APznzaalselifJKjGQdTCA51cF7bldYdFMvDcshM8opKFZ_ZaIV-dqkiLoIKIfhz2tS6Fw5UBk25u...
 
A Survey on Sentiment Analysis and Opinion Mining.pdf
A Survey on Sentiment Analysis and Opinion Mining.pdfA Survey on Sentiment Analysis and Opinion Mining.pdf
A Survey on Sentiment Analysis and Opinion Mining.pdf
 
A Survey On Sentiment Analysis And Opinion Mining Techniques
A Survey On Sentiment Analysis And Opinion Mining TechniquesA Survey On Sentiment Analysis And Opinion Mining Techniques
A Survey On Sentiment Analysis And Opinion Mining Techniques
 
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
 
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
XAI LANGUAGE TUTOR - A XAI-BASED LANGUAGE LEARNING CHATBOT USING ONTOLOGY AND...
 
[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...
[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...
[PhDThesis2021] - Augmenting the knowledge pyramid with unconventional data a...
 
Discussing about information architecture for language learning: apps and str...
Discussing about information architecture for language learning: apps and str...Discussing about information architecture for language learning: apps and str...
Discussing about information architecture for language learning: apps and str...
 
Leb08talksept17
Leb08talksept17Leb08talksept17
Leb08talksept17
 
NLP unit-VI.pptx
NLP unit-VI.pptxNLP unit-VI.pptx
NLP unit-VI.pptx
 
Crowdsourcing Documentation in Software Engineering
Crowdsourcing Documentation in Software EngineeringCrowdsourcing Documentation in Software Engineering
Crowdsourcing Documentation in Software Engineering
 
New Research Articles 2020 May Issue International Journal of Software Engin...
New Research Articles 2020 May  Issue International Journal of Software Engin...New Research Articles 2020 May  Issue International Journal of Software Engin...
New Research Articles 2020 May Issue International Journal of Software Engin...
 

More from Martino Mensio

Towards a Cross-Article Narrative Comparison of News
Towards a Cross-Article Narrative Comparison of NewsTowards a Cross-Article Narrative Comparison of News
Towards a Cross-Article Narrative Comparison of NewsMartino Mensio
 
Detecting subtle text manipulations
Detecting subtle text manipulationsDetecting subtle text manipulations
Detecting subtle text manipulationsMartino Mensio
 
News Source Credibility in the Eyes of Different Assessors
News Source Credibility in the Eyes of Different AssessorsNews Source Credibility in the Eyes of Different Assessors
News Source Credibility in the Eyes of Different AssessorsMartino Mensio
 
A Multi-layer LSTM-based Approach for Robot Command Interaction Modeling
A Multi-layer LSTM-based Approach for Robot Command Interaction ModelingA Multi-layer LSTM-based Approach for Robot Command Interaction Modeling
A Multi-layer LSTM-based Approach for Robot Command Interaction ModelingMartino Mensio
 
The Rise of Emotion-aware Conversational Agents: Threats in Digital Emotions
The Rise of Emotion-aware Conversational Agents: Threats in Digital EmotionsThe Rise of Emotion-aware Conversational Agents: Threats in Digital Emotions
The Rise of Emotion-aware Conversational Agents: Threats in Digital EmotionsMartino Mensio
 
Deep Learning per la Comprensione del Linguaggio Naturale - HKN
Deep Learning per la Comprensione del Linguaggio Naturale - HKNDeep Learning per la Comprensione del Linguaggio Naturale - HKN
Deep Learning per la Comprensione del Linguaggio Naturale - HKNMartino Mensio
 

More from Martino Mensio (6)

Towards a Cross-Article Narrative Comparison of News
Towards a Cross-Article Narrative Comparison of NewsTowards a Cross-Article Narrative Comparison of News
Towards a Cross-Article Narrative Comparison of News
 
Detecting subtle text manipulations
Detecting subtle text manipulationsDetecting subtle text manipulations
Detecting subtle text manipulations
 
News Source Credibility in the Eyes of Different Assessors
News Source Credibility in the Eyes of Different AssessorsNews Source Credibility in the Eyes of Different Assessors
News Source Credibility in the Eyes of Different Assessors
 
A Multi-layer LSTM-based Approach for Robot Command Interaction Modeling
A Multi-layer LSTM-based Approach for Robot Command Interaction ModelingA Multi-layer LSTM-based Approach for Robot Command Interaction Modeling
A Multi-layer LSTM-based Approach for Robot Command Interaction Modeling
 
The Rise of Emotion-aware Conversational Agents: Threats in Digital Emotions
The Rise of Emotion-aware Conversational Agents: Threats in Digital EmotionsThe Rise of Emotion-aware Conversational Agents: Threats in Digital Emotions
The Rise of Emotion-aware Conversational Agents: Threats in Digital Emotions
 
Deep Learning per la Comprensione del Linguaggio Naturale - HKN
Deep Learning per la Comprensione del Linguaggio Naturale - HKNDeep Learning per la Comprensione del Linguaggio Naturale - HKN
Deep Learning per la Comprensione del Linguaggio Naturale - HKN
 

Recently uploaded

Analytical Profile of Coleus Forskohlii | Forskolin .pptx
Analytical Profile of Coleus Forskohlii | Forskolin .pptxAnalytical Profile of Coleus Forskohlii | Forskolin .pptx
Analytical Profile of Coleus Forskohlii | Forskolin .pptxSwapnil Therkar
 
Work, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE PhysicsWork, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE Physicsvishikhakeshava1
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Patrick Diehl
 
Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Nistarini College, Purulia (W.B) India
 
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...Sérgio Sacani
 
Boyles law module in the grade 10 science
Boyles law module in the grade 10 scienceBoyles law module in the grade 10 science
Boyles law module in the grade 10 sciencefloriejanemacaya1
 
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsSérgio Sacani
 
zoogeography of pakistan.pptx fauna of Pakistan
zoogeography of pakistan.pptx fauna of Pakistanzoogeography of pakistan.pptx fauna of Pakistan
zoogeography of pakistan.pptx fauna of Pakistanzohaibmir069
 
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaDashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaPraksha3
 
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝soniya singh
 
Luciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptxLuciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptxAleenaTreesaSaji
 
Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...
Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...
Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...Sérgio Sacani
 
Recombination DNA Technology (Microinjection)
Recombination DNA Technology (Microinjection)Recombination DNA Technology (Microinjection)
Recombination DNA Technology (Microinjection)Jshifa
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...RohitNehra6
 
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfAnalytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfSwapnil Therkar
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxUmerFayaz5
 
Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )aarthirajkumar25
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real timeSatoshi NAKAHIRA
 
Physiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptxPhysiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptxAArockiyaNisha
 

Recently uploaded (20)

Analytical Profile of Coleus Forskohlii | Forskolin .pptx
Analytical Profile of Coleus Forskohlii | Forskolin .pptxAnalytical Profile of Coleus Forskohlii | Forskolin .pptx
Analytical Profile of Coleus Forskohlii | Forskolin .pptx
 
Work, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE PhysicsWork, Energy and Power for class 10 ICSE Physics
Work, Energy and Power for class 10 ICSE Physics
 
Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?Is RISC-V ready for HPC workload? Maybe?
Is RISC-V ready for HPC workload? Maybe?
 
The Philosophy of Science
The Philosophy of ScienceThe Philosophy of Science
The Philosophy of Science
 
Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...Bentham & Hooker's Classification. along with the merits and demerits of the ...
Bentham & Hooker's Classification. along with the merits and demerits of the ...
 
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
 
Boyles law module in the grade 10 science
Boyles law module in the grade 10 scienceBoyles law module in the grade 10 science
Boyles law module in the grade 10 science
 
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
 
zoogeography of pakistan.pptx fauna of Pakistan
zoogeography of pakistan.pptx fauna of Pakistanzoogeography of pakistan.pptx fauna of Pakistan
zoogeography of pakistan.pptx fauna of Pakistan
 
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tantaDashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
Dashanga agada a formulation of Agada tantra dealt in 3 Rd year bams agada tanta
 
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
Call Girls in Munirka Delhi 💯Call Us 🔝8264348440🔝
 
Luciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptxLuciferase in rDNA technology (biotechnology).pptx
Luciferase in rDNA technology (biotechnology).pptx
 
Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...
Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...
Discovery of an Accretion Streamer and a Slow Wide-angle Outflow around FUOri...
 
Recombination DNA Technology (Microinjection)
Recombination DNA Technology (Microinjection)Recombination DNA Technology (Microinjection)
Recombination DNA Technology (Microinjection)
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...
 
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdfAnalytical Profile of Coleus Forskohlii | Forskolin .pdf
Analytical Profile of Coleus Forskohlii | Forskolin .pdf
 
Animal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptxAnimal Communication- Auditory and Visual.pptx
Animal Communication- Auditory and Visual.pptx
 
Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )
 
Grafana in space: Monitoring Japan's SLIM moon lander in real time
Grafana in space: Monitoring Japan's SLIM moon lander  in real timeGrafana in space: Monitoring Japan's SLIM moon lander  in real time
Grafana in space: Monitoring Japan's SLIM moon lander in real time
 
Physiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptxPhysiochemical properties of nanomaterials and its nanotoxicity.pptx
Physiochemical properties of nanomaterials and its nanotoxicity.pptx
 

Deep Semantic Learning for Conversational Agents

  • 1. Deep Semantic Learning for Conversational Agents Candidate: Martino Mensio Supervisor: Maurizio Morisio Tutor: Giuseppe Rizzo 12 April 2018 1
  • 2. Objectives 1. Identify the approaches to build a Conversational Agent with Natural Language Understanding 2. Use the context of interaction 2
  • 4. Background: Conversational Agents What they can do: - automated interaction with customer - virtual assistants What content they can provide: - Chit-chat (small talk) - Goal-oriented - Knowledge-based 4
  • 6. Background: an example of Understanding 6
  • 8. Background: intent classification [1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association. 8
  • 9. Background: slot filling [1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association. 9
  • 10. Background: Word Embeddings [2] Harris, Z. S. (1970). Distributional structure. In Papers in structural and transformational linguistics (pp. 775-794). Springer, Dordrecht. 10 Distributional Semantics [2]: words used in similar contexts have similar meaning - each word corresponds to a vector of reals - small dimensionality (50~300) - semantic distribution in a multidimensional space
  • 12. Approach: the multi-turn interactions - detect the change of intent - capture intent dependencies - consider the agent words 12
  • 13. Approach: difference between multi-turn and single-turn 13
  • 15. Approach: Word Embeddings for Italian language recomputation of Italian Wikipedia embeddings with proper tokenization (with respect to [6]) 15[6] Berardi, G., Esuli, A., & Marcheggiani, D. (2015). Word Embeddings Go to Italy: A Comparison of Models and Training Datasets. In IIR. “Voglio una bici vicino a piazza castello, grazie” ↓ [“Voglio”, “una”, “bici”, “vicino”, “a”, “piazza”, “castello”, “,”, “grazie”]
  • 17. Results: the datasets available: - ATIS (single-turn) [3] - nlu-benchmark (single-turn) [4] - kvret (multi-turn) [5] collected: - bikes Italian (single-turn) - bikes English (single-turn) 17 [3] Hemphill, C., Godfrey, J., Doddington, G. (1990). The ATIS spoken language systems pilot corpus. DARPA Speech and Natural Language Workshop [4] https://github.com/snipsco/nlu-benchmark [5] Eric, M. and Manning, C. (2017). Key-value retrieval networks for task-oriented dialogue. SIGDIAL 2017: Session on Natural Language Generation for Dialog Systems
  • 18. Results: multi-turn intent classification results on kvret dataset [5] 18 approach F1 epoch number intent RNN agent words ✓ LSTM ✓ 0.9987 7 ✓ LSTM ✘ 0.9987 8 ✓ GRU ✓ 0.9975 14 ✘ ✓ 0.9951 5 ✓ GRU ✘ 0.9585 9 [1]✘ ✘ 0.8524 8 [1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association. [5] Eric, M. and Manning, C. (2017). Key-value retrieval networks for task-oriented dialogue. SIGDIAL 2017: Session on Natural Language Generation for Dialog Systems
  • 19. Results: Italian Word Embeddings 19 [7] Mikolov, T., Yih, W. T., & Zweig, G. (2013). Linguistic regularities in continuous space word representations. In Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (pp. 746-751). Word Embeddings accuracy Italian values from [6] on Wikipedia 44.81% Computed Italian values on Wikipedia 58.14% analogy test [7]: - semantic (capital-country, nationality adjective, currency, family) - syntactic (m-f, singular-plural, tenses, comparatives, superlatives) [6] Berardi, G., Esuli, A., & Marcheggiani, D. (2015). Word Embeddings Go to Italy: A Comparison of Models and Training Datasets. In IIR.
  • 20. Results: the difference on the global tasks (Italian) Measured on the bike sharing dataset on the approach by [1] 20 Word Embeddings intent classification F1 slot filling F1 Italian values from [6] on Wikipedia, 730k vectors 0.8421 0.5666 Computed Italian values on Wikipedia, 758k vectors 0.8947 0.6153 [7] Berardi, G., Esuli, A., & Marcheggiani, D. (2015). Word Embeddings Go to Italy: A Comparison of Models and Training Datasets. In IIR. [1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association.
  • 21. Results: the difference of embeddings on the two tasks (English) 21 Embeddings intent classification F1 slot filling F1 ATIS nlu-bench mark bikes english ATIS nlu-bench mark bikes english Trainable, random initialization 0.9740 0.9928 0.9428 0.9425 0.9177 0.9000 [8] precomputed, 685k keys, 20k unique vectors 0.9660 0.9928 0.9714 0.9588 0.8970 0.9375 [8] precomputed, 685k keys, 685k unique vectors 0.9860 0.9928 0.9714 0.9649 0.9170 0.9689 [1] Liu, B. and Lane, I. (2016). Attention-based recurrent neural network models for joint intent detection and slot filling. Proceedings of The 17th Annual Meeting of the International Speech Communication Association. [8] https://spacy.io/models/en Measured on the approach by [1]
  • 22. Conclusions - results of the multi-turn show the importance of context - results for the word embeddings show the importance of their proper choice 22
  • 23. Future works - multi-turn slot filling to remove handcrafted dialog tracking 23