SlideShare ist ein Scribd-Unternehmen logo
1 von 26
QAestro - Semantic-based Composition of
Question Answering Pipeline
DEXA 2017
28.08.2017
Kuldeep Singh, Ioanna Lytra, Maria Esther Vidal, Dharmen Punjani,
Harsh Thakkar, Christoph Lange, Sören Auer
Motivating Example
● Question: “Where was Albert Einstein born?”
● Query in a Formal Language: e.g. “SPARQL Query”
2
Motivating Example
● Question: “Where was Albert Einstein born?”
● Query in a Formal Language: e.g. “SPARQL Query”
3
Motivating Example
4
Question Answering Tasks
● More than 70 QA
systems in last 5
years.
● QA systems
implement similar
tasks to answer
user’s question.
5
Motivating Example
Challenge
6
OKBQA Question Answering Framework
Problem Statement
● Define a framework able to
○ semantically describe QA components and QA developer
requirements; and
○ produce QA component compositions based on these
semantic descriptions.
7
Our Contributions
● The QAestro framework that generates QA component
compositions based on developer requirements.
● A vocabulary for expressing QA tasks and developer
requirements.
● The formalization of existing 51 QA components from 20 QA
systems.
● A mapping of the QA component composition problem into
Query Rewriting Problem (QRP).
● An empirical evaluation of QAestro framework.
8
Agenda
1. Approach
a. Controlled Vocabulary for Abstract QA Tasks
b. Semantic Descriptions of QA Component
c. QA Developer Requirement
d. The Problem of QA Component Composition
2. Empirical Study
3. Conclusions
9
Approach
Controlled Vocabulary for Abstract QA Tasks
● QAV= (𝛿,A)
○ 𝛿: signature of logical language.
○ A: set of axioms describing the relationships among vocabulary
concepts.
11
Controlled Vocabulary for Abstract QA Tasks
● QAV= (𝛿,A)
○ 𝛿: signature of logical language.
○ A: set of axioms describing the relationships among vocabulary
concepts.
● Example: Disambiguation(x, y, z, t) → QuestionAnalysis(x, y)
○ where x=entity; y=question; z=disambiguated entities, t=template ;
12
Semantic Description of QA Component
● Follow Local as View (LAV) approach to define QA components.
○ Allows easy scaling up to large number of QA components.
● Example:
○ Describe Stanford NER that performs named entity recognition
task.
13
Semantic Description of QA Component
● Example:
○ StanfordNER($y,x) :– Recognition(y,x), Question(y), Entity(x)
■ Recognition is the QA task of Stanford NER component.
■ StanfordNER accepts question as input and gives recognised
entities as output.
■ Question, Entity, variables x and y are part of controlled
vocabulary.
14
QA Developer Requirement
● Example: Give me all the components that implements entity
recognition and entity disambiguation task.
● Semantic Description of QA Developer Requirement:
○ QADevReq($y, x) :– Recognition(y,x), Disambiguation(x,y,z,t)
15
The Problem of QA Component Composition
● How to compose valid compositions using LAV mappings of QA
component.
LAV Mapping Examples:
○ DBpediaNER($y, x) :– Recognition(y, x), Question(y), Entity(x)
○ Alchemy($y, z) :– Disambiguation(x, y, z, t), Question(y), DisEntity(z)
○ Qakisatype($y, a) :– Answertype(y, a, o), Question(y), Atype(a)
○ StanfordNER($y,x) :– Recognition(y,x), Question(y), Entity(x)
○ Agdistis($x, $y, z) :– Disambiguation(x, y, z, t), Entity(x), Question(y),
disEntity(z)
16
17
QAestro Framework
QA Component Composition using QAestro
● QADevReq($y, x) :– Recognition(y, x), Disambiguation(x, y, z, t),
answertype(y, a, o)
Results:
○ QADevReq($y, x) :– StanfordNER($y, x), Agdistis($y, $x, z),
Qakisatype($y, a)
○ QADevReq($q, e) :– DBpediaNER($y, x), Agdistis($y, $x, z),
Qakisatype($y, a)
18
Empirical Study
Experiment Configuration
● QAestro Implementation
○ Implemented in python 2.7 on top of MCDSAT.
○ Source Code: https://github.com/WDAqua/QAestro
○ 51 QA component formalisation for 20 QA systems
● QAestro Experiments
○ Executed on a laptop (with Fedora Linux 25) Intel i7-4550U,
4x1.50GHz and 8GB RAM.
20
Analysis of QA Components
21
Analysis of QA Components
22
QA Component Composition
23
QA Component Composition
24
Conclusions
● Introduce QAestro- a framework that:
○ Semantically describe QA Components and Developer
Requirements.
○ Follows Local as View approach.
○ Compose valid compositions of QA components.
○ Can successfully deal with the growing number of QA.
○ Demonstrates efficient processing time.
25
Find us!
● For more information please visit: http://wdaqua.eu/QAestro/
● Demo can be viewed at:
○ https://www.youtube.com/watch?v=9lhamebx7JM&feature
=youtu.be
● Write me on : kuldeep.singh@iais.fraunhofer.de
26

Weitere ähnliche Inhalte

Ähnlich wie QAestro semantic based composition of QA pipelines

Question Answering with Subgraph Embeddings
Question Answering with Subgraph EmbeddingsQuestion Answering with Subgraph Embeddings
Question Answering with Subgraph EmbeddingsKarel Ha
 
Extensible Operators and Literals for JavaScript
Extensible Operators and Literals for JavaScriptExtensible Operators and Literals for JavaScript
Extensible Operators and Literals for JavaScriptBrendan Eich
 
R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...
R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...
R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...Edureka!
 
Mca se chapter_9_formal_methods
Mca se chapter_9_formal_methodsMca se chapter_9_formal_methods
Mca se chapter_9_formal_methodsAman Adhikari
 
PyCon Ukraine 2016: Maintaining a high load Python project for newcomers
PyCon Ukraine 2016: Maintaining a high load Python project for newcomersPyCon Ukraine 2016: Maintaining a high load Python project for newcomers
PyCon Ukraine 2016: Maintaining a high load Python project for newcomersViach Kakovskyi
 
Sql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPrague
Sql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPragueSql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPrague
Sql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPragueLuis Beltran
 
WISS QA Do it yourself Question answering over Linked Data
WISS QA Do it yourself Question answering over Linked DataWISS QA Do it yourself Question answering over Linked Data
WISS QA Do it yourself Question answering over Linked DataAndre Freitas
 
Web Page Ranking using Machine Learning
Web Page Ranking using Machine LearningWeb Page Ranking using Machine Learning
Web Page Ranking using Machine LearningPradip Rahul
 
Der Test Manager ist tot - lang lebe der Test Master
Der Test Manager ist tot - lang lebe der Test MasterDer Test Manager ist tot - lang lebe der Test Master
Der Test Manager ist tot - lang lebe der Test MasterSwissQ Consulting AG
 
Andrea Lattuada, Gabriele Petronella - Building startups on Scala
Andrea Lattuada, Gabriele Petronella - Building startups on ScalaAndrea Lattuada, Gabriele Petronella - Building startups on Scala
Andrea Lattuada, Gabriele Petronella - Building startups on ScalaScala Italy
 
AutoML for Data Science Productivity and Toward Better Digital Decisions
AutoML for Data Science Productivity and Toward Better Digital DecisionsAutoML for Data Science Productivity and Toward Better Digital Decisions
AutoML for Data Science Productivity and Toward Better Digital DecisionsSteven Gustafson
 
Recommending job ads to people
Recommending job ads to peopleRecommending job ads to people
Recommending job ads to peopleFabian Abel
 
Using Deep Learning and Customized Solr Components to Improve search Relevanc...
Using Deep Learning and Customized Solr Components to Improve search Relevanc...Using Deep Learning and Customized Solr Components to Improve search Relevanc...
Using Deep Learning and Customized Solr Components to Improve search Relevanc...Lucidworks
 
Predicting SPARQL query execution time and suggesting SPARQL queries based on...
Predicting SPARQL query execution time and suggesting SPARQL queries based on...Predicting SPARQL query execution time and suggesting SPARQL queries based on...
Predicting SPARQL query execution time and suggesting SPARQL queries based on...Rakebul Hasan
 

Ähnlich wie QAestro semantic based composition of QA pipelines (20)

Question Answering with Subgraph Embeddings
Question Answering with Subgraph EmbeddingsQuestion Answering with Subgraph Embeddings
Question Answering with Subgraph Embeddings
 
Extensible Operators and Literals for JavaScript
Extensible Operators and Literals for JavaScriptExtensible Operators and Literals for JavaScript
Extensible Operators and Literals for JavaScript
 
R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...
R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...
R Programming For Beginners | R Language Tutorial | R Tutorial For Beginners ...
 
Mca se chapter_9_formal_methods
Mca se chapter_9_formal_methodsMca se chapter_9_formal_methods
Mca se chapter_9_formal_methods
 
PyCon Ukraine 2016: Maintaining a high load Python project for newcomers
PyCon Ukraine 2016: Maintaining a high load Python project for newcomersPyCon Ukraine 2016: Maintaining a high load Python project for newcomers
PyCon Ukraine 2016: Maintaining a high load Python project for newcomers
 
Sql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPrague
Sql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPragueSql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPrague
Sql Server Machine Learning Services - Sql Saturday Prague 2018 #SqlSatPrague
 
STRICT-SANER2017
STRICT-SANER2017STRICT-SANER2017
STRICT-SANER2017
 
WISS QA Do it yourself Question answering over Linked Data
WISS QA Do it yourself Question answering over Linked DataWISS QA Do it yourself Question answering over Linked Data
WISS QA Do it yourself Question answering over Linked Data
 
Web Page Ranking using Machine Learning
Web Page Ranking using Machine LearningWeb Page Ranking using Machine Learning
Web Page Ranking using Machine Learning
 
Der Test Manager ist tot - lang lebe der Test Master
Der Test Manager ist tot - lang lebe der Test MasterDer Test Manager ist tot - lang lebe der Test Master
Der Test Manager ist tot - lang lebe der Test Master
 
Sam zhang demo
Sam zhang demoSam zhang demo
Sam zhang demo
 
Step By Step Guide to Learn R
Step By Step Guide to Learn RStep By Step Guide to Learn R
Step By Step Guide to Learn R
 
Andrea Lattuada, Gabriele Petronella - Building startups on Scala
Andrea Lattuada, Gabriele Petronella - Building startups on ScalaAndrea Lattuada, Gabriele Petronella - Building startups on Scala
Andrea Lattuada, Gabriele Petronella - Building startups on Scala
 
AutoML for Data Science Productivity and Toward Better Digital Decisions
AutoML for Data Science Productivity and Toward Better Digital DecisionsAutoML for Data Science Productivity and Toward Better Digital Decisions
AutoML for Data Science Productivity and Toward Better Digital Decisions
 
Recommending job ads to people
Recommending job ads to peopleRecommending job ads to people
Recommending job ads to people
 
Using Deep Learning and Customized Solr Components to Improve search Relevanc...
Using Deep Learning and Customized Solr Components to Improve search Relevanc...Using Deep Learning and Customized Solr Components to Improve search Relevanc...
Using Deep Learning and Customized Solr Components to Improve search Relevanc...
 
Mini_Project
Mini_ProjectMini_Project
Mini_Project
 
SamKK
SamKKSamKK
SamKK
 
Query compiler
Query compilerQuery compiler
Query compiler
 
Predicting SPARQL query execution time and suggesting SPARQL queries based on...
Predicting SPARQL query execution time and suggesting SPARQL queries based on...Predicting SPARQL query execution time and suggesting SPARQL queries based on...
Predicting SPARQL query execution time and suggesting SPARQL queries based on...
 

Kürzlich hochgeladen

The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024Rafal Los
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...Neo4j
 
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsMaria Levchenko
 
Developing An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of BrazilDeveloping An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of BrazilV3cube
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘RTylerCroy
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationRadu Cotescu
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking MenDelhi Call girls
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesSinan KOZAK
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Igalia
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...Martijn de Jong
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...apidays
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)Gabriella Davis
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Enterprise Knowledge
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Miguel Araújo
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountPuma Security, LLC
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processorsdebabhi2
 
Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Paola De la Torre
 
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of ServiceCNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Servicegiselly40
 

Kürzlich hochgeladen (20)

The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...Workshop - Best of Both Worlds_ Combine  KG and Vector search for  enhanced R...
Workshop - Best of Both Worlds_ Combine KG and Vector search for enhanced R...
 
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
Neo4j - How KGs are shaping the future of Generative AI at AWS Summit London ...
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
Developing An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of BrazilDeveloping An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of Brazil
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men08448380779 Call Girls In Greater Kailash - I Women Seeking Men
08448380779 Call Girls In Greater Kailash - I Women Seeking Men
 
Unblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen FramesUnblocking The Main Thread Solving ANRs and Frozen Frames
Unblocking The Main Thread Solving ANRs and Frozen Frames
 
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
 
A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)A Domino Admins Adventures (Engage 2024)
A Domino Admins Adventures (Engage 2024)
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
Breaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path MountBreaking the Kubernetes Kill Chain: Host Path Mount
Breaking the Kubernetes Kill Chain: Host Path Mount
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101Salesforce Community Group Quito, Salesforce 101
Salesforce Community Group Quito, Salesforce 101
 
CNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of ServiceCNv6 Instructor Chapter 6 Quality of Service
CNv6 Instructor Chapter 6 Quality of Service
 

QAestro semantic based composition of QA pipelines

  • 1. QAestro - Semantic-based Composition of Question Answering Pipeline DEXA 2017 28.08.2017 Kuldeep Singh, Ioanna Lytra, Maria Esther Vidal, Dharmen Punjani, Harsh Thakkar, Christoph Lange, Sören Auer
  • 2. Motivating Example ● Question: “Where was Albert Einstein born?” ● Query in a Formal Language: e.g. “SPARQL Query” 2
  • 3. Motivating Example ● Question: “Where was Albert Einstein born?” ● Query in a Formal Language: e.g. “SPARQL Query” 3
  • 5. ● More than 70 QA systems in last 5 years. ● QA systems implement similar tasks to answer user’s question. 5 Motivating Example
  • 7. Problem Statement ● Define a framework able to ○ semantically describe QA components and QA developer requirements; and ○ produce QA component compositions based on these semantic descriptions. 7
  • 8. Our Contributions ● The QAestro framework that generates QA component compositions based on developer requirements. ● A vocabulary for expressing QA tasks and developer requirements. ● The formalization of existing 51 QA components from 20 QA systems. ● A mapping of the QA component composition problem into Query Rewriting Problem (QRP). ● An empirical evaluation of QAestro framework. 8
  • 9. Agenda 1. Approach a. Controlled Vocabulary for Abstract QA Tasks b. Semantic Descriptions of QA Component c. QA Developer Requirement d. The Problem of QA Component Composition 2. Empirical Study 3. Conclusions 9
  • 11. Controlled Vocabulary for Abstract QA Tasks ● QAV= (𝛿,A) ○ 𝛿: signature of logical language. ○ A: set of axioms describing the relationships among vocabulary concepts. 11
  • 12. Controlled Vocabulary for Abstract QA Tasks ● QAV= (𝛿,A) ○ 𝛿: signature of logical language. ○ A: set of axioms describing the relationships among vocabulary concepts. ● Example: Disambiguation(x, y, z, t) → QuestionAnalysis(x, y) ○ where x=entity; y=question; z=disambiguated entities, t=template ; 12
  • 13. Semantic Description of QA Component ● Follow Local as View (LAV) approach to define QA components. ○ Allows easy scaling up to large number of QA components. ● Example: ○ Describe Stanford NER that performs named entity recognition task. 13
  • 14. Semantic Description of QA Component ● Example: ○ StanfordNER($y,x) :– Recognition(y,x), Question(y), Entity(x) ■ Recognition is the QA task of Stanford NER component. ■ StanfordNER accepts question as input and gives recognised entities as output. ■ Question, Entity, variables x and y are part of controlled vocabulary. 14
  • 15. QA Developer Requirement ● Example: Give me all the components that implements entity recognition and entity disambiguation task. ● Semantic Description of QA Developer Requirement: ○ QADevReq($y, x) :– Recognition(y,x), Disambiguation(x,y,z,t) 15
  • 16. The Problem of QA Component Composition ● How to compose valid compositions using LAV mappings of QA component. LAV Mapping Examples: ○ DBpediaNER($y, x) :– Recognition(y, x), Question(y), Entity(x) ○ Alchemy($y, z) :– Disambiguation(x, y, z, t), Question(y), DisEntity(z) ○ Qakisatype($y, a) :– Answertype(y, a, o), Question(y), Atype(a) ○ StanfordNER($y,x) :– Recognition(y,x), Question(y), Entity(x) ○ Agdistis($x, $y, z) :– Disambiguation(x, y, z, t), Entity(x), Question(y), disEntity(z) 16
  • 18. QA Component Composition using QAestro ● QADevReq($y, x) :– Recognition(y, x), Disambiguation(x, y, z, t), answertype(y, a, o) Results: ○ QADevReq($y, x) :– StanfordNER($y, x), Agdistis($y, $x, z), Qakisatype($y, a) ○ QADevReq($q, e) :– DBpediaNER($y, x), Agdistis($y, $x, z), Qakisatype($y, a) 18
  • 20. Experiment Configuration ● QAestro Implementation ○ Implemented in python 2.7 on top of MCDSAT. ○ Source Code: https://github.com/WDAqua/QAestro ○ 51 QA component formalisation for 20 QA systems ● QAestro Experiments ○ Executed on a laptop (with Fedora Linux 25) Intel i7-4550U, 4x1.50GHz and 8GB RAM. 20
  • 21. Analysis of QA Components 21
  • 22. Analysis of QA Components 22
  • 25. Conclusions ● Introduce QAestro- a framework that: ○ Semantically describe QA Components and Developer Requirements. ○ Follows Local as View approach. ○ Compose valid compositions of QA components. ○ Can successfully deal with the growing number of QA. ○ Demonstrates efficient processing time. 25
  • 26. Find us! ● For more information please visit: http://wdaqua.eu/QAestro/ ● Demo can be viewed at: ○ https://www.youtube.com/watch?v=9lhamebx7JM&feature =youtu.be ● Write me on : kuldeep.singh@iais.fraunhofer.de 26

Hinweis der Redaktion

  1. In this slide, I will talk about the problem of QA component orchestrations in a framework.
  2. A signature is a set of predicate and constant symbols, from which logical formulas can be constructed, whereas the axioms A describe the vocabulary by illustrating the relationships between concepts.
  3. A signature is a set of predicate and constant symbols, from which logical formulas can be constructed, whereas the axioms A describe the vocabulary by illustrating the relationships between concepts. disambiguation is a predicate of arity four in ; disambig(x; y; z; t) denotes that the QA task . disambiguation relates an entity x, a question y, a disambiguated entity z, and a template t. Furthermore, the binary predicate questionAnalysis(x; y) models the question analysis task and relates an entity x to a question y.
  4. QA components are defined formally w.r.t their functionality, input and output dependencies Each component is defined as conjunctive rule considering All the variables in the head of a rule are also variables in the predicates in the body of the rule.
  5. QA components are defined formally w.r.t their functionality, input and output dependencies Each component is defined as conjunctive rule considering All the variables in the head of a rule are also variables in the predicates in the body of the rule.
  6. Talk about Stanford NER has its input y i.e. question
  7. Talk about Stanford NER has its input y i.e. question
  8. Talk about Stanford NER has its input y i.e. question
  9. that in almost half of the QA systems, components that implement Tokenization and Query Generation are included, while some less popular QA tasks like Answer Type Identification and Syntactic Parser are part of only two QA systems.
  10. Talk about Stanford NER has its input y i.e. question
  11. Talk about Stanford NER has its input y i.e. question