A Semantic Search Approach to Task-Completion EnginesDarío Garigliotti
Date: July 8, 2018
Venue: Ann Arbor, MI, USA. Doctoral Consortium at the 41st International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '18)
Please cite, link to or credit this presentation when using it or part of it in your work.
Introduction to Enterprise Search. A two hour class to introduce Enterprise Search. It covers:
The problems enterprise search can solve
History of (web) search
How we search and find?
Current state of Enterprise Search + stats
Technical concept
Information quality
Feedback cycle
Five dimensions of Findability
Date: March 3rd, 2016
Venue: Trondheim, Norway. Doctoral Seminar at NTNU
Please cite, link to or credit this presentation when using it or part of it in your work.
1. SharePoint 2010 introduces a new Managed Metadata Service that allows for centralized storage and management of terms across sites and site collections. This provides a consistent way to organize content.
2. The Managed Metadata Service supports both taxonomies for structured terms as well as folksonomies for user-generated keywords and tags. It integrates with other features like Business Connectivity Services.
3. While powerful, the Managed Metadata Service requires planning to set up terms and administer the term store. Considerations include importing structures metadata, separating terms with commas, and preventing misspellings.
This tutorial gives an overview of how search engines and machine learning techniques can be tightly coupled to address the need for building scalable recommender or other prediction based systems. Typically, most of them architect retrieval and prediction in two phases. In Phase I, a search engine returns the top-k results based on constraints expressed as a query. In Phase II, the top-k results are re-ranked in another system according to an optimization function that uses a supervised trained model. However this approach presents several issues, such as the possibility of returning sub-optimal results due to the top-k limits during query, as well as the prescence of some inefficiencies in the system due to the decoupling of retrieval and ranking.
To address this issue the authors created ML-Scoring, an open source framework that tightly integrates machine learning models into Elasticsearch, a popular search engine. ML-Scoring replaces the default information retrieval ranking function with a custom supervised model that is trained through Spark, Weka, or R that is loaded as a plugin in Elasticsearch. This tutorial will not only review basic methods in information retrieval and machine learning, but it will also walk through practical examples from loading a dataset into Elasticsearch to training a model in Spark, Weka, or R, to creating the ML-Scoring plugin for Elasticsearch. No prior experience is required in any system listed (Elasticsearch, Spark, Weka, R), though some programming experience is recommended.
RecSys 2015 Tutorial – Scalable Recommender Systems: Where Machine Learning...S. Diana Hu
Search engines have focused on solving the document retrieval problem, so their scoring functions do not handle naturally non-traditional IR data types, such as numerical or categorical. Therefore, on domains beyond traditional search, scores representing strengths of associations or matches may vary widely. As such, the original model doesn’t suffice, so relevance ranking is performed as a two-phase approach with 1) regular search 2) external model to re-rank the filtered items. Metrics such as click-through and conversion rates are associated with the users’ response to items served. The predicted selection rates that arise in real-time can be critical for optimal matching. For example, in recommender systems, predicted performance of a recommended item in a given context, also called response prediction, is often used in determining a set of recommendations to serve in relation to a given serving opportunity. Similar techniques are used in the advertising domain. To address this issue the authors have created ML-Scoring, an open source framework that tightly integrates machine learning models into a popular search engine (SOLR/Elasticsearch), replacing the default IR-based ranking function. A custom model is trained through either Weka or Spark and it is loaded as a plugin used at query time to compute custom scores.
Broad introduction to information retrieval and web search, used to teaching at the Yahoo Bangalore Summer School 2013. Slides are a mash-up from my own and other people's presentations.
A Beginner's Guide to Machine Learning with Scikit-LearnSarah Guido
Given at the PyData NYC 2013 conference (http://vimeo.com/79517341), and will be given at PyTennessee 2014.
Scikit-learn is one of the most well-known machine learning Python modules in existence. But how does it work, and what, for that matter, is machine learning? For those with programming experience but who are new to machine learning, this talk gives a beginner-level overview of how machine learning can be useful, important machine learning concepts, and how to implement them with scikit-learn. We’ll use real world data to look at supervised and unsupervised machine learning algorithms and why scikit-learn is useful for performing these tasks.
A Semantic Search Approach to Task-Completion EnginesDarío Garigliotti
Date: July 8, 2018
Venue: Ann Arbor, MI, USA. Doctoral Consortium at the 41st International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR '18)
Please cite, link to or credit this presentation when using it or part of it in your work.
Introduction to Enterprise Search. A two hour class to introduce Enterprise Search. It covers:
The problems enterprise search can solve
History of (web) search
How we search and find?
Current state of Enterprise Search + stats
Technical concept
Information quality
Feedback cycle
Five dimensions of Findability
Date: March 3rd, 2016
Venue: Trondheim, Norway. Doctoral Seminar at NTNU
Please cite, link to or credit this presentation when using it or part of it in your work.
1. SharePoint 2010 introduces a new Managed Metadata Service that allows for centralized storage and management of terms across sites and site collections. This provides a consistent way to organize content.
2. The Managed Metadata Service supports both taxonomies for structured terms as well as folksonomies for user-generated keywords and tags. It integrates with other features like Business Connectivity Services.
3. While powerful, the Managed Metadata Service requires planning to set up terms and administer the term store. Considerations include importing structures metadata, separating terms with commas, and preventing misspellings.
This tutorial gives an overview of how search engines and machine learning techniques can be tightly coupled to address the need for building scalable recommender or other prediction based systems. Typically, most of them architect retrieval and prediction in two phases. In Phase I, a search engine returns the top-k results based on constraints expressed as a query. In Phase II, the top-k results are re-ranked in another system according to an optimization function that uses a supervised trained model. However this approach presents several issues, such as the possibility of returning sub-optimal results due to the top-k limits during query, as well as the prescence of some inefficiencies in the system due to the decoupling of retrieval and ranking.
To address this issue the authors created ML-Scoring, an open source framework that tightly integrates machine learning models into Elasticsearch, a popular search engine. ML-Scoring replaces the default information retrieval ranking function with a custom supervised model that is trained through Spark, Weka, or R that is loaded as a plugin in Elasticsearch. This tutorial will not only review basic methods in information retrieval and machine learning, but it will also walk through practical examples from loading a dataset into Elasticsearch to training a model in Spark, Weka, or R, to creating the ML-Scoring plugin for Elasticsearch. No prior experience is required in any system listed (Elasticsearch, Spark, Weka, R), though some programming experience is recommended.
RecSys 2015 Tutorial – Scalable Recommender Systems: Where Machine Learning...S. Diana Hu
Search engines have focused on solving the document retrieval problem, so their scoring functions do not handle naturally non-traditional IR data types, such as numerical or categorical. Therefore, on domains beyond traditional search, scores representing strengths of associations or matches may vary widely. As such, the original model doesn’t suffice, so relevance ranking is performed as a two-phase approach with 1) regular search 2) external model to re-rank the filtered items. Metrics such as click-through and conversion rates are associated with the users’ response to items served. The predicted selection rates that arise in real-time can be critical for optimal matching. For example, in recommender systems, predicted performance of a recommended item in a given context, also called response prediction, is often used in determining a set of recommendations to serve in relation to a given serving opportunity. Similar techniques are used in the advertising domain. To address this issue the authors have created ML-Scoring, an open source framework that tightly integrates machine learning models into a popular search engine (SOLR/Elasticsearch), replacing the default IR-based ranking function. A custom model is trained through either Weka or Spark and it is loaded as a plugin used at query time to compute custom scores.
Broad introduction to information retrieval and web search, used to teaching at the Yahoo Bangalore Summer School 2013. Slides are a mash-up from my own and other people's presentations.
A Beginner's Guide to Machine Learning with Scikit-LearnSarah Guido
Given at the PyData NYC 2013 conference (http://vimeo.com/79517341), and will be given at PyTennessee 2014.
Scikit-learn is one of the most well-known machine learning Python modules in existence. But how does it work, and what, for that matter, is machine learning? For those with programming experience but who are new to machine learning, this talk gives a beginner-level overview of how machine learning can be useful, important machine learning concepts, and how to implement them with scikit-learn. We’ll use real world data to look at supervised and unsupervised machine learning algorithms and why scikit-learn is useful for performing these tasks.
Introduction to enterprise search for intranets and websitesKristian Norling
An introduction to Enterprise Search. A two hour course to introduce Enterprise Search by Kristian Norling. This is a class I love to do, so if you have interest in it for on premise/in-house class or at a conference or such, please contact me.
The course covers:
- Problems for enterprise search to solve.
- Web Search
- How we search and find?
- Current state of Enterprise Search, including stats
- Technical concept
- Information quality and metadata
- Feedback cycle
- Five dimensions of Findability
STATIK is a systems thinking approach to implementing Kanban developed by Mike Burrows. It is a repeatable process with 6 steps: 1) understand sources of dissatisfaction, 2) analyze demand and capability, 3) model the knowledge discovery process, 4) discover classes of service, 5) design Kanban systems, and 6) roll out. The document discusses each step in STATIK and how it can be used both to initially implement Kanban and to reinvigorate existing Kanban implementations.
This document provides an overview of a course on data warehousing, filtering, and mining. The course is being taught in Fall 2004 at Temple University. The document includes the course syllabus which outlines topics like data warehousing, OLAP technology, data preprocessing, mining association rules, classification, cluster analysis, and mining complex data types. Grading will be based on assignments, quizzes, a presentation, individual project, and final exam. The document also provides introductory material on data mining including definitions and examples.
The document provides an overview of EPiServer Find, an advanced search engine for EPiServer. It discusses unified search capabilities including searching across different content types and customizing projections. It also covers highlighting search results, usage statistics and tracking, and demo functionality like autocomplete, spellchecking, and did you mean suggestions. The presentation includes demos of filtering, facets, and multi-search queries. Lab exercises are proposed to practice common search scenarios using the EPiServer Find API.
The document summarizes research on exploring the relationship between reflective writing and information literacy development in undergraduate business students. Key points:
- Students completed a group project and individual reflections assessing their information literacy and contribution to the project.
- Reflections were analyzed using the SCONUL 7 Pillars model of information literacy and models of reflective practice.
- Preliminary findings show students reflected at various depths on different information literacy concepts mapped in the 7 Pillars, with deeper reflection on searching skills than identifying information needs.
Data mining Basics and complete description onwordSulman Ahmed
This document discusses data mining and provides examples of its applications. It begins by explaining why data is mined from both commercial and scientific viewpoints in order to discover useful patterns and information. It then discusses some of the challenges of data mining, such as dealing with large datasets, high dimensionality, complex data types, and distributed data sources. The document outlines common data mining tasks like classification, clustering, association rule mining, and regression. It provides real-world examples of how these techniques are used for applications like fraud detection, customer profiling, and scientific discovery.
- The document discusses setting up an effective content management system in SharePoint by developing a content architecture and taxonomy. It covers key concepts like content types, site columns, and metadata that form the building blocks of organizing content in SharePoint.
- An effective content architecture relies on defining relevant content types and site columns and associating them with terms from the taxonomy at the appropriate levels to properly categorize and surface content.
- The presenter provides guidance on how to strategically design content types and site columns that align with business needs and allow content to be consistently organized across sites.
BASPUG May 2014 - Taming Your Taxonomy in SharePointJonathan Ralton
This document provides an agenda and overview for a presentation on taming taxonomies in SharePoint. The presentation covers content architecture and taxonomy concepts, metadata such as content types and site columns, and best practices for implementing metadata in SharePoint. It discusses defining the appropriate scope and hierarchy for content types and columns. The goal is to help attendees understand how metadata supports findability and usability of content in SharePoint.
Getting Started With User Research, Presented at Agile2010Carol Smith
This document provides an overview of conducting user research within an Agile development process. It discusses quick and inexpensive research methods like interviews, observations, card sorting and usability testing that can be integrated into Sprints. The goal is to understand user needs and behaviors to effectively share information with the team. It emphasizes starting research by identifying primary and secondary user groups and their tasks, goals and environments. The document recommends iterative user research and testing of prototypes to help focus efforts and get to an 80% understanding of users.
How to Do Research: Seven Steps to Successful ResearchSi Krishan
This document provides guidance on how to conduct research. It discusses what research is, different knowledge sources, and a seven step process for successful research: defining the research question, gathering information, forming a hypothesis, implementation/analysis, organizing results, communicating results, and revising. It also covers defining a good research question, reading papers, organizing references, and the nonlinear nature of research. The overall document serves as a guide for students and researchers on how to effectively plan and execute a research project.
Modern Search: Using ML & NLP advances to enhance search and discoveryAll Things Open
Presented at Open Source Charlotte
Presented by Grant Ingersoll
Title: Modern Search: Using ML & NLP advances to enhance search and discovery
Abstract: With the recent advances in natural language processing and machine learning thanks to deep learning and large general purpose models, many search applications are confronted with how best to upgrade their systems, if at all. In this talk, we’ll look at practical ways to enhance search using neural and other machine learning techniques across ranking, content understanding and query understanding. We’ll also look at the tradeoffs of traditional approaches with a goal of helping you decide what’s best for your application.
For more info on Open Source Charlotte: https://www.meetup.com/open-source-charlotte/
Consumer research consists of systematically collecting and analyzing data to aid decision makers in developing goods, services, and ideas. There are two main types of research: qualitative exploratory research that attempts to understand a phenomenon and provide initial insights, and quantitative conclusive research that confirms preliminary insights and informs appropriate actions. The research process involves defining objectives, designing the study, collecting data through various methods like surveys and experiments, analyzing the results, and presenting findings. Ethics require research to maintain confidentiality, be unbiased, and relevant to its context.
The document discusses contextualized online search and research skills. It covers topics like search tools, information evaluation, and plagiarism. It provides guidance on using search engines effectively through search operators and techniques. It emphasizes the importance of evaluating information sources for accuracy, authority, objectivity and currency. Examples of information sources discussed include indigenous knowledge, libraries and the internet. The document aims to help students improve their ability to conduct credible online research.
Personalizing Content Using Taxonomy with Megan Gilhooly, Vice President Cust...LavaConConference
Watch the recording! - https://youtu.be/8P8LMgcaZpg
Technical content is playing an increasingly important role in the overall digital experience for leading companies. The goal of your content must be to provide relevant, personalized answers to technical questions about your product as quickly as possible. This lets you unlock the true value and ROI in your technical content resources.
Megan will demonstrate:
What is a taxonomy
How a taxonomy helps you facilitate excellence in personalized content delivery
The basics of taxonomy design for technical content
Exploratory Search upon Semantically Described Web Data Sources: Service regi...Marco Brambilla
This presentation was given at the SSW workshop, collocated with VLDB 2012.
Exploratory search applications upon structured Web content is becoming one of the main information seeking paradigms for users. This is mainly due to the move towards mobile and pervasive Web access and to the more and more tight intertwining between everyday life and information seeking.
Structured data is typically distributed on the Web and accessible through a service-oriented paradigm. This paper proposes a vision on: (1) a semantically-enabled service registration framework for describing in a Web data services in a convenient way; and (2) a design method for applications that exploit such model using a design pattern -based method.
The document discusses a presentation on taming taxonomies in SharePoint. It covers content architecture and taxonomy concepts in theory, and explores content types, site columns, and metadata in practice. The presentation includes exercises to design content structures and apply metadata using SharePoint's building blocks.
We provide real time big data training in Chennai by industrial experts with real time scenarios.
Our Advanced topics will enhance the students expectations into high level knowledge in Big Data Technology.
For More Info.Reach our Big Data Technical Team@ +91 96677211551/56
The Experience of Big data Training Experts Team.
www.thecreatingexperts.com
SAP BEST INSTITUTES IN CHENNAI
http://www.youtube.com/watch?v=UpWthI0P-7g
This document provides an introduction to big data and basic data analysis. It discusses the large amounts of data being generated every day from sources like the web, social networks, and scientific projects. It also introduces some common data types and how data is stored and analyzed. Key concepts covered include relational and non-relational data, data warehousing using star schemas, online analytical processing, data mining techniques like classification and clustering, and working with data streams. The document aims to give an overview of the big data landscape and basic analytical methods.
The document discusses spend classification and provides definitions and challenges related to spend classification including categorization at source, inconsistent categorization, multiple disparate taxonomies, and classifying spend into miscellaneous categories. It also discusses taxonomy, standard vs custom taxonomies, and how machine learning such as named entity recognition can help with spend classification.
Date: March 22, 2019
Venue: Stavanger, Norway. Symposium at the IAI group
Please cite, link to or credit this presentation when using it or part of it in your work.
About "Towards Better Text Understanding and Retrieval through Kernel Entity ...Darío Garigliotti
Summary of the paper "Towards Better Text Understanding and Retrieval through Kernel Entity Salience Modeling", presented at SIGIR 2018.
Date: October 17, 2018
Venue: London, UK. Reading group
Please cite, link to or credit this presentation when using it or part of it in your work.
Introduction to enterprise search for intranets and websitesKristian Norling
An introduction to Enterprise Search. A two hour course to introduce Enterprise Search by Kristian Norling. This is a class I love to do, so if you have interest in it for on premise/in-house class or at a conference or such, please contact me.
The course covers:
- Problems for enterprise search to solve.
- Web Search
- How we search and find?
- Current state of Enterprise Search, including stats
- Technical concept
- Information quality and metadata
- Feedback cycle
- Five dimensions of Findability
STATIK is a systems thinking approach to implementing Kanban developed by Mike Burrows. It is a repeatable process with 6 steps: 1) understand sources of dissatisfaction, 2) analyze demand and capability, 3) model the knowledge discovery process, 4) discover classes of service, 5) design Kanban systems, and 6) roll out. The document discusses each step in STATIK and how it can be used both to initially implement Kanban and to reinvigorate existing Kanban implementations.
This document provides an overview of a course on data warehousing, filtering, and mining. The course is being taught in Fall 2004 at Temple University. The document includes the course syllabus which outlines topics like data warehousing, OLAP technology, data preprocessing, mining association rules, classification, cluster analysis, and mining complex data types. Grading will be based on assignments, quizzes, a presentation, individual project, and final exam. The document also provides introductory material on data mining including definitions and examples.
The document provides an overview of EPiServer Find, an advanced search engine for EPiServer. It discusses unified search capabilities including searching across different content types and customizing projections. It also covers highlighting search results, usage statistics and tracking, and demo functionality like autocomplete, spellchecking, and did you mean suggestions. The presentation includes demos of filtering, facets, and multi-search queries. Lab exercises are proposed to practice common search scenarios using the EPiServer Find API.
The document summarizes research on exploring the relationship between reflective writing and information literacy development in undergraduate business students. Key points:
- Students completed a group project and individual reflections assessing their information literacy and contribution to the project.
- Reflections were analyzed using the SCONUL 7 Pillars model of information literacy and models of reflective practice.
- Preliminary findings show students reflected at various depths on different information literacy concepts mapped in the 7 Pillars, with deeper reflection on searching skills than identifying information needs.
Data mining Basics and complete description onwordSulman Ahmed
This document discusses data mining and provides examples of its applications. It begins by explaining why data is mined from both commercial and scientific viewpoints in order to discover useful patterns and information. It then discusses some of the challenges of data mining, such as dealing with large datasets, high dimensionality, complex data types, and distributed data sources. The document outlines common data mining tasks like classification, clustering, association rule mining, and regression. It provides real-world examples of how these techniques are used for applications like fraud detection, customer profiling, and scientific discovery.
- The document discusses setting up an effective content management system in SharePoint by developing a content architecture and taxonomy. It covers key concepts like content types, site columns, and metadata that form the building blocks of organizing content in SharePoint.
- An effective content architecture relies on defining relevant content types and site columns and associating them with terms from the taxonomy at the appropriate levels to properly categorize and surface content.
- The presenter provides guidance on how to strategically design content types and site columns that align with business needs and allow content to be consistently organized across sites.
BASPUG May 2014 - Taming Your Taxonomy in SharePointJonathan Ralton
This document provides an agenda and overview for a presentation on taming taxonomies in SharePoint. The presentation covers content architecture and taxonomy concepts, metadata such as content types and site columns, and best practices for implementing metadata in SharePoint. It discusses defining the appropriate scope and hierarchy for content types and columns. The goal is to help attendees understand how metadata supports findability and usability of content in SharePoint.
Getting Started With User Research, Presented at Agile2010Carol Smith
This document provides an overview of conducting user research within an Agile development process. It discusses quick and inexpensive research methods like interviews, observations, card sorting and usability testing that can be integrated into Sprints. The goal is to understand user needs and behaviors to effectively share information with the team. It emphasizes starting research by identifying primary and secondary user groups and their tasks, goals and environments. The document recommends iterative user research and testing of prototypes to help focus efforts and get to an 80% understanding of users.
How to Do Research: Seven Steps to Successful ResearchSi Krishan
This document provides guidance on how to conduct research. It discusses what research is, different knowledge sources, and a seven step process for successful research: defining the research question, gathering information, forming a hypothesis, implementation/analysis, organizing results, communicating results, and revising. It also covers defining a good research question, reading papers, organizing references, and the nonlinear nature of research. The overall document serves as a guide for students and researchers on how to effectively plan and execute a research project.
Modern Search: Using ML & NLP advances to enhance search and discoveryAll Things Open
Presented at Open Source Charlotte
Presented by Grant Ingersoll
Title: Modern Search: Using ML & NLP advances to enhance search and discovery
Abstract: With the recent advances in natural language processing and machine learning thanks to deep learning and large general purpose models, many search applications are confronted with how best to upgrade their systems, if at all. In this talk, we’ll look at practical ways to enhance search using neural and other machine learning techniques across ranking, content understanding and query understanding. We’ll also look at the tradeoffs of traditional approaches with a goal of helping you decide what’s best for your application.
For more info on Open Source Charlotte: https://www.meetup.com/open-source-charlotte/
Consumer research consists of systematically collecting and analyzing data to aid decision makers in developing goods, services, and ideas. There are two main types of research: qualitative exploratory research that attempts to understand a phenomenon and provide initial insights, and quantitative conclusive research that confirms preliminary insights and informs appropriate actions. The research process involves defining objectives, designing the study, collecting data through various methods like surveys and experiments, analyzing the results, and presenting findings. Ethics require research to maintain confidentiality, be unbiased, and relevant to its context.
The document discusses contextualized online search and research skills. It covers topics like search tools, information evaluation, and plagiarism. It provides guidance on using search engines effectively through search operators and techniques. It emphasizes the importance of evaluating information sources for accuracy, authority, objectivity and currency. Examples of information sources discussed include indigenous knowledge, libraries and the internet. The document aims to help students improve their ability to conduct credible online research.
Personalizing Content Using Taxonomy with Megan Gilhooly, Vice President Cust...LavaConConference
Watch the recording! - https://youtu.be/8P8LMgcaZpg
Technical content is playing an increasingly important role in the overall digital experience for leading companies. The goal of your content must be to provide relevant, personalized answers to technical questions about your product as quickly as possible. This lets you unlock the true value and ROI in your technical content resources.
Megan will demonstrate:
What is a taxonomy
How a taxonomy helps you facilitate excellence in personalized content delivery
The basics of taxonomy design for technical content
Exploratory Search upon Semantically Described Web Data Sources: Service regi...Marco Brambilla
This presentation was given at the SSW workshop, collocated with VLDB 2012.
Exploratory search applications upon structured Web content is becoming one of the main information seeking paradigms for users. This is mainly due to the move towards mobile and pervasive Web access and to the more and more tight intertwining between everyday life and information seeking.
Structured data is typically distributed on the Web and accessible through a service-oriented paradigm. This paper proposes a vision on: (1) a semantically-enabled service registration framework for describing in a Web data services in a convenient way; and (2) a design method for applications that exploit such model using a design pattern -based method.
The document discusses a presentation on taming taxonomies in SharePoint. It covers content architecture and taxonomy concepts in theory, and explores content types, site columns, and metadata in practice. The presentation includes exercises to design content structures and apply metadata using SharePoint's building blocks.
We provide real time big data training in Chennai by industrial experts with real time scenarios.
Our Advanced topics will enhance the students expectations into high level knowledge in Big Data Technology.
For More Info.Reach our Big Data Technical Team@ +91 96677211551/56
The Experience of Big data Training Experts Team.
www.thecreatingexperts.com
SAP BEST INSTITUTES IN CHENNAI
http://www.youtube.com/watch?v=UpWthI0P-7g
This document provides an introduction to big data and basic data analysis. It discusses the large amounts of data being generated every day from sources like the web, social networks, and scientific projects. It also introduces some common data types and how data is stored and analyzed. Key concepts covered include relational and non-relational data, data warehousing using star schemas, online analytical processing, data mining techniques like classification and clustering, and working with data streams. The document aims to give an overview of the big data landscape and basic analytical methods.
The document discusses spend classification and provides definitions and challenges related to spend classification including categorization at source, inconsistent categorization, multiple disparate taxonomies, and classifying spend into miscellaneous categories. It also discusses taxonomy, standard vs custom taxonomies, and how machine learning such as named entity recognition can help with spend classification.
Similar to Task-Based Support in Search Engines (20)
Date: March 22, 2019
Venue: Stavanger, Norway. Symposium at the IAI group
Please cite, link to or credit this presentation when using it or part of it in your work.
About "Towards Better Text Understanding and Retrieval through Kernel Entity ...Darío Garigliotti
Summary of the paper "Towards Better Text Understanding and Retrieval through Kernel Entity Salience Modeling", presented at SIGIR 2018.
Date: October 17, 2018
Venue: London, UK. Reading group
Please cite, link to or credit this presentation when using it or part of it in your work.
Highlights of the 40th European Conference on Information Retrieval (ECIR '18)
Date: April 6, 2018
Venue: Stavanger, Norway. Symposium at the IAI group
Please cite, link to or credit this presentation when using it or part of it in your work.
A Semantic Search Approach to Task-Completion EnginesDarío Garigliotti
Date: February 27, 2018
Venue: Stavanger, Norway. UiS TN910 - Innovation and Project Awareness
Please cite, link to or credit this presentation when using it or part of it in your work.
This document summarizes Darío Garigliotti's work on constructing a knowledge base of entity-oriented search intents. It introduces key concepts like entities, entity types, RDF tuples, and knowledge bases. It then describes a pipeline approach for building the knowledge base, which involves acquiring refiners from queries, categorizing refiners, discovering intents, and constructing the knowledge base with triples linking intents to entities, categories, and expressing refiners. Evaluation is done on the accuracy of the extracted knowledge base facts. The full knowledge base contains 155k triples describing 31k intent profiles across 581 entity types. Potential applications include leveraging the knowledge base to identify intents in new queries and improving entity cards.
Date: October 2nd, 2017
Venue: Amsterdam, The Netherlands. The 2017 ACM SIGIR International Conference on Theory of Information Retrieval (ICTIR '17)
Corresponding article: https://arxiv.org/abs/1708.08291
Please cite, link to or credit this presentation when using it or part of it in your work.
Learning-to-Rank Target Types for Entity-Bearing QueriesDarío Garigliotti
Date: October 1st, 2017
Venue: Amsterdam, The Netherlands. LEARNER 2017, co-located with the 2017 ACM SIGIR International Conference on Theory of Information Retrieval (ICTIR '17)
Corresponding article: http://ceur-ws.org/Vol-2007/LEARNER2017_short_3.pdf
Please cite, link to or credit this presentation when using it or part of it in your work.
Date: March 13, 2017
Venue: Stavanger, Norway. Doctoral Seminar at the IAI group for the research visit of Prof. Maarten de Rijke
Please cite, link to or credit this presentation when using it or part of it in your work.
Date: October 7, 2016
Venue: Stavanger, Norway. Technical talk at UiS TN-IDE
Please cite, link to or credit this presentation when using it or part of it in your work.
Date: June 14, 2016
Venue: Oslo, Norway. Doctoral Seminar at HiOA
Please cite, link to or credit this presentation when using it or part of it in your work.
Date: June 10, 2016
Venue: Stavanger, Norway. Doctoral Seminar at the IAI group for the research visit of Prof. Kalervo Järvelin
Please cite, link to or credit this presentation when using it or part of it in your work.
Date: March 4, 2016
Venue: Trondheim, Norway. Doctoral Seminar at NTNU
Please cite, link to or credit this presentation when using it or part of it in your work.
Original title in Spanish: Si ésta es la respuesta, ¿cuál era la pregunta?
Date: November 20, 2013
Venue: Córdoba, Argentina. Project on Question Generation for the MSc Specialization Course "Natural Language Processing" (Faculty of Mathematics, Astronomy, Physics and Computation, National University of Córdoba)
Please cite, link to or credit this presentation when using it or part of it in your work.
Semi-supervised Learning for Word Sense DisambiguationDarío Garigliotti
Original title in Spanish: Desambiguación de Palabras Polisémicas mediante Aprendizaje Semi-supervisado
Date: September 20, 2013
Venue: Córdoba, Argentina. 42nd JAIIO - Argentine Journals of Informatics and Operating Research (JAIIO '13)
Please cite, link to or credit this presentation when using it or part of it in your work.
Semi-supervised Learning for Word Sense DisambiguationDarío Garigliotti
Original title in Spanish: Desambiguación de Palabras Polisémicas mediante Aprendizaje Semi-supervisado
Date: November 19, 2012
Venue: Córdoba, Argentina. Project on Word Sense Disambiguation for the MSc Specialization Course "Artificial Intelligence" at FaMAF, UNC (Faculty of Mathematics, Astronomy, Physics and Computation, National University of Córdoba)
Video: https://www.youtube.com/watch?v=qv9qZaBw-Qw
Date: August 2016
Venue: Saratov, Russian Federation. The 10th Russian Summer School in Information Retrieval (RuSSIR '16)
Please cite, link to or credit this presentation when using it or part of it in your work.
Semi-supervised Learning for Word Sense DisambiguationDarío Garigliotti
Original title in Spanish: Desambiguación de Palabras Polisémicas mediante Aprendizaje Semi-supervisado
Date: September 2013
Venue: Córdoba, Argentina. 42nd JAIIO - Argentine Journals of Informatics and Operating Research (JAIIO '13)
Corresponding article: https://arxiv.org/abs/1908.09641
Please cite the paper, and link to or credit this presentation when using it or part of it in your work.
Hierarchical clustering builds clusters hierarchically, by either merging or splitting clusters at each step. Agglomerative hierarchical clustering starts with each point as a separate cluster and successively merges the closest clusters based on a defined proximity measure between clusters. This results in a dendrogram showing the nested clustering structure. The basic algorithm computes a proximity matrix, then repeatedly merges the closest pair of clusters and updates the matrix until all points are in one cluster.
The document discusses several alternative classification techniques including rule-based classifiers, nearest neighbors classifiers, and Naive Bayes classifiers. It provides examples of how each technique works and some key aspects to consider, such as how to build rule-based classifiers directly from data or indirectly from other models like decision trees. It also covers concepts like mutual exclusivity of rules, rule coverage and accuracy, and how to order rules.
ESR spectroscopy in liquid food and beverages.pptxPRIYANKA PATEL
With increasing population, people need to rely on packaged food stuffs. Packaging of food materials requires the preservation of food. There are various methods for the treatment of food to preserve them and irradiation treatment of food is one of them. It is the most common and the most harmless method for the food preservation as it does not alter the necessary micronutrients of food materials. Although irradiated food doesn’t cause any harm to the human health but still the quality assessment of food is required to provide consumers with necessary information about the food. ESR spectroscopy is the most sophisticated way to investigate the quality of the food and the free radicals induced during the processing of the food. ESR spin trapping technique is useful for the detection of highly unstable radicals in the food. The antioxidant capability of liquid food and beverages in mainly performed by spin trapping technique.
ESA/ACT Science Coffee: Diego Blas - Gravitational wave detection with orbita...Advanced-Concepts-Team
Presentation in the Science Coffee of the Advanced Concepts Team of the European Space Agency on the 07.06.2024.
Speaker: Diego Blas (IFAE/ICREA)
Title: Gravitational wave detection with orbital motion of Moon and artificial
Abstract:
In this talk I will describe some recent ideas to find gravitational waves from supermassive black holes or of primordial origin by studying their secular effect on the orbital motion of the Moon or satellites that are laser ranged.
The technology uses reclaimed CO₂ as the dyeing medium in a closed loop process. When pressurized, CO₂ becomes supercritical (SC-CO₂). In this state CO₂ has a very high solvent power, allowing the dye to dissolve easily.
Current Ms word generated power point presentation covers major details about the micronuclei test. It's significance and assays to conduct it. It is used to detect the micronuclei formation inside the cells of nearly every multicellular organism. It's formation takes place during chromosomal sepration at metaphase.
(June 12, 2024) Webinar: Development of PET theranostics targeting the molecu...Scintica Instrumentation
Targeting Hsp90 and its pathogen Orthologs with Tethered Inhibitors as a Diagnostic and Therapeutic Strategy for cancer and infectious diseases with Dr. Timothy Haystead.
Mending Clothing to Support Sustainable Fashion_CIMaR 2024.pdfSelcen Ozturkcan
Ozturkcan, S., Berndt, A., & Angelakis, A. (2024). Mending clothing to support sustainable fashion. Presented at the 31st Annual Conference by the Consortium for International Marketing Research (CIMaR), 10-13 Jun 2024, University of Gävle, Sweden.
7. Motivation
• Today's web search experience aims to understand
the user query
• A way to organize information is via structured
knowledge centered around entities
• Large knowledge repositories and knowledge bases
have become available
7
15. • Underlying search goal is often a complex and
knowledge-intensive task
• For example, to plan a travel
- How to get there?
- Where to stay?
- What to do?
• Task completion would provide support for the
user when accomplishing complex search
tasks
Motivation
15
18. An example
• Searching for wedding cakes
cake shops
Konditoriet i
Sandnes
Olja's Kake
Boutique
Baker
Corner Lura
18
19. An example
• Searching for wedding cakes
wedding cake shops
Stavanger
Conditori
Olja's Kake
Boutique
Gjestalveien
Conditori
19
20. • Entity retrieval is the task of obtaining a
ranked list of entities relevant to a search
query
• We investigate the utilization of entity type
information for entity retrieval
Entity Retrieval
20
23. Type-aware Entity Retrieval
• Type information is known to improve entity
retrieval
• Yet it is a multifaceted problem
query entity
wedding cake shops
target types
Stavanger Conditori
term-based
similarity
type-based
similarity
… …
entity types
23
24. How can entity type information
be utilized in ad-hoc entity retrieval?
24
25. • We assume oracle-given type information
Type-aware Entity Retrieval
…
Shop
… ……
Company
Organization …
cake shopscake shops
Bank
Law Firm …
25
26. • We assume oracle-given type information
• We identify dimensions in utilizing entity
type information
- Type taxonomy
- Type representation
- Retrieval model
Type-aware Entity Retrieval
26
27. • Which type taxonomy to use?
- DBpedia Ontology (7 levels, 600 types)
- Freebase Types (2 levels, 2K types)
- Wikipedia Categories (34 levels, 600K types)
- YAGO Taxonomy (19 levels, 500K types)
• These vary a lot in terms of hierarchical
structure and in how entity-type assignments
are recorded
Type-aware Entity Retrieval
27
28. • How to represent the hierarchical information?
Type-aware Entity Retrieval
28
29. • How to use type information into entity
retrieval?
• Retrieval task is defined in a generative
probabilistic framework P(q | e)
• Both query and entity are considered
in the term space as well as in the type
space
Type-aware Entity Retrieval
29
33. • We assume oracle-given type information
• We conduct an evaluation of dimensions in
utilizing entity type information
- Type taxonomy
- Type representation
- Retrieval model
• We use a strong text-based baseline
• We test with the DBpedia Entity collection v2
Type-aware Entity Retrieval
33
34. • Wikipedia, in combination with the most
specific type representation, performs best
• Hierarchical relationships from ancestor
types improve retrieval effectiveness, but
most specific types provide the best
performance
• Results regarding most effective type-aware
retrieval model vary across configurations
Type-aware Entity Retrieval
34
39. Target Type Identification
• "We assume oracle-given type information"
- How to identify target entity types?
- How do these target types automatically
identified perform for type-aware entity
retrieval?
39
40. How can we automatically
identify target entity types?
40
41. Target Type Identification
• We revisit the task of hierarchical target
type identification
• Task: to find the main target types of a
query, from a type taxonomy, such that
these are the most specific category of
entities that are relevant to the query.
If no matching type can be found in the
taxonomy then the query is assigned a
special NIL-type
41
42. Target Type Identification
• We develop a Learning-to-Rank approach
• We evaluate it using a purpose-built test
collection
42
44. • We also conduct an evaluation utilizing,
rather than a target types oracle, target
entity types automatically identified
44
Automatic types for Entity
Retrieval
46. We identify and evaluate dimensions in utilizing
target entity type information for ad-hoc entity
retrieval.
We build a test collection for target entity type
identification, and develop and evaluate a
Learning-to-Rank approach for this problem.
SUMMARY
46
49. Entity-Oriented Search
Intents
• Intent: the underlying user need in a entity-
oriented search query
- For example, the intent of booking a hotel room
• Refiner: a way to express an intent in an
entity-oriented query
- For example, for booking a hotel room:
"booking", "book", "reservation", "rooms"
49
62. <fashion designer> instagram
Understanding entity-
oriented search intents
• We obtain a collection of type-level query patterns
stella mccartney instagram
vivienne westwood instagram
62
63. Understanding entity-
oriented search intents
• We obtain a collection of type-level query patterns
• Pick a Freebase type if it covers 100+ prominent entities
• Get query suggestions for top 1000- entities per type
• For each query, replace entity by type
• Aggregate all frequencies for each (type, refiner) pair
• Filter out all type-level refiners with frequency of 4-
• Select 50 representative types by stratified sampling
63
69. Understanding entity-
oriented search intents
• We define a scheme of intent categories
- Website, Property, Service, Other
=> Website
=> Property
=> Service
vivienne westwood age
vivienne westwood instagram
vivienne westwood customer care
69
70. Understanding entity-
oriented search intents
• We annotate 2.3K+ unique type-level refiners
with intent category via crowdsourcing
• We observe the proportions of refiners in each
category
Property: 28.6%
Service: 54.06%
Website: 5.34%
Other: 12.08%
70
71. Understanding entity-
oriented search intents
71
organization
business operation
chemical compound
film
location
event
food
hotel
disease
restaurant
travel destination
0
50
100
150
200
250
university
house
person
newspaper
airport
basketball player
album
professional sports team
game
artwork
0
50
100
150
200
railway
human language
tv station
political party
amusement park
exhibition venue
chef
programming language
academic institution
netflix genre
0
20
40
60
80
100
120
war
currency
blogger
hobby
football match
sports championship
star
muscle
olympic sport
company
0
10
20
30
40
50
WroSicaO cycOone
kingdoP
PedicaO sSeciaOWy
coPic book SubOisher
oiO fieOd
Wower
beer counWry region
eOecWion
asWeroid
beOief
0
10
20
30
40
50
3roSerWy
WebsiWe
Service
2Wher
72. We propose a scheme of entity-oriented search
intent categories.
We annotate a collection of query refiners using
the scheme, and observe that there is a large
proportion of service-oriented intents.
SUMMARY
72
74. How can we build a knowledge base
of entity-oriented search intents?
74
75. 1. Intents searched for a type of entities
paris map, sydney map => [city] map
2. Categories assigned to refiners
vivienne westwood instagram => Website
vivienne westwood age => Property
vivienne westwood customer care => Service
3. Multiple refiners expressing an intent
"booking", "book", "make a reservation", "rooms"
75
A knowledge base of entity-
oriented search intents
76. 1. Intents searched for a type of entities
paris map, sydney map => [city] map
• (intent ID, searchedForType, entity type, confidence)
2. Categories assigned to refiners
vivienne westwood instagram => Website
vivienne westwood age => Property
vivienne westwood customer care => Service
3. Multiple refiners expressing an intent
"booking", "book", "make a reservation", "rooms"
76
A knowledge base of entity-
oriented search intents
77. 1. Intents searched for a type of entities
paris map, sydney map => [city] map
• (intent ID, searchedForType, entity type, confidence)
2. Categories assigned to refiners
vivienne westwood instagram => Website
vivienne westwood age => Property
vivienne westwood customer care => Service
• (intent ID, ofCategory, intent category, confidence)
3. Multiple refiners expressing an intent
"booking", "book", "make a reservation", "rooms"
77
A knowledge base of entity-
oriented search intents
78. 1. Intents searched for a type of entities
paris map, sydney map => [city] map
• (intent ID, searchedForType, entity type, confidence)
2. Categories assigned to refiners
vivienne westwood instagram => Website
vivienne westwood age => Property
vivienne westwood customer care => Service
• (intent ID, ofCategory, intent category, confidence)
3. Multiple refiners expressing an intent
"booking", "book", "make a reservation", "rooms"
• (intent ID, expressedBy, refiner, confidence)
A knowledge base of entity-
oriented search intents
78
79. Approach
Refiners
acquisition
Refiners
categorization
Intents
discovery
[hotel] airport
[hotel] spa
[hotel] booking
...
[hotel] airport: Service
[hotel] address: Property
[hotel] expedia: Website
...
taxi
arrive
Hotel_Arrivingbooking
make a reservation
Hotel_Booking
address
Hotel_Address
KB
construction
Intent ID Predicate Object Confidence
Hotel_Booking searchedForType [hotel] c1
Hotel_Booking ofCategory Service c2
Hotel_Booking expressedBy "booking" c3
Hotel_Booking expressedBy "make a reservation" c4
Hotel_Booking expressedBy "rooms" c5
79
80. Approach
Refiners
acquisition
Refiners
categorization
Intents
discovery
[hotel] airport
[hotel] spa
[hotel] booking
...
[hotel] airport: Service
[hotel] address: Property
[hotel] expedia: Website
...
taxi
arrive
Hotel_Arrivingbooking
make a reservation
Hotel_Booking
address
Hotel_Address
Intent
profile
{ KB
construction
Intent ID Predicate Object Confidence
Hotel_Booking searchedForType [hotel] c1
Hotel_Booking ofCategory Service c2
Hotel_Booking expressedBy "booking" c3
Hotel_Booking expressedBy "make a reservation" c4
Hotel_Booking expressedBy "rooms" c5
80
81. Knowledge base construction
• Application of the pipeline to extract all
quadruples from 581 unseen types
• 155K quadruples, 31K intent profiles
- Excerpt of the KB, for intent ID
<aviation.airline-65-customer_service>
81
82. Experimental evaluation
• Experts judge correctness, ignoring
confidence, of around 1.29% of IntentsKB
82
[0, 0.87) [0.87, 0.88) [0.88, 0.9) [0.9, 0.93) [0.93, 1]
Confidence intervals according to the splitting percentiles
0%
20%
40%
60%
80%
100%
Proportionoftriples
6,337 6,370 6,335 6,368 6,314
Correct
Incorrect, OFCATEGORY
Incorrect, EXPRESSEDBY
83. We design and build a knowledge base of entity-
oriented search intents.
We evaluate each component in our approach,
as well as the correctness of the obtained
knowledge base.
SUMMARY
83
93. Cheap wedding
cake
Make your own invitations
Buy a used wedding gownExcerpt from TREC Tasks
test dataset
low wedding budget
1 low budget wedding dresses
0 low wedding budget cars
1 find a gown
...
0 wedding flowers
1 cup cake wedding
1 wedding cakes
...
2 wedding invitation
1 find wedding invitation templates
0 designer dresses wedding
...
An example
93
94. Cheap wedding
cake
Make your own invitations
Buy a used wedding gownExcerpt from TREC Tasks
test dataset
}
low wedding budget
1 low budget wedding dresses
0 low wedding budget cars
1 find a gown
...
0 wedding flowers
1 cup cake wedding
1 wedding cakes
...
2 wedding invitation
1 find wedding invitation templates
0 designer dresses wedding
...
An example
94
95. Cheap wedding
cake
Make your own invitations
Buy a used wedding gownExcerpt from TREC Tasks
test dataset
}
}
low wedding budget
1 low budget wedding dresses
0 low wedding budget cars
1 find a gown
...
0 wedding flowers
1 cup cake wedding
1 wedding cakes
...
2 wedding invitation
1 find wedding invitation templates
0 designer dresses wedding
...
An example
95
96. An example
Cheap wedding
cake
Make your own invitations
Buy a used wedding gownExcerpt from TREC Tasks
test dataset
}
}
}
low wedding budget
1 low budget wedding dresses
0 low wedding budget cars
1 find a gown
...
0 wedding flowers
1 cup cake wedding
1 wedding cakes
...
2 wedding invitation
1 find wedding invitation templates
0 designer dresses wedding
...
96
97. How can we generate query suggestions
for supporting task-based search?
97
98. • Given an initial query,
Suggesting queries to
support task-based search
wedding cake
wedding cake gallery
wedding cake recipes
wedding cake flavors
98
99. • Given an initial query,
to get a ranked list of
query suggestions
that cover all the
possible subtasks
related to the task
that the user is trying
to achieve.
Suggesting queries to
support task-based search
wedding cake
wedding cake gallery
wedding cake recipes
wedding cake flavors
99
100. • Given an initial query,
to get a ranked list of
query suggestions
that cover all the
possible subtasks
related to the task
that the user is trying
to achieve.
Suggesting queries to
support task-based search
wedding cake
wedding cake gallery
wedding cake recipes
wedding cake flavors
• This is the task
understanding
problem
100
101. Suggesting queries to
support task-based search
• We propose an end-to-end generative
probabilistic model
• We exploit different information sources
101
106. • Components:
• Source importance
q0
API SUGGS. WEB SNIPPETS WEB DOCS. WH
Suggesting queries to
support task-based search
• We propose an end-to-end generative
probabilistic model
106
107. • Components:
• Source importance
• Document importance
q0
API SUGGS. WEB SNIPPETS WEB DOCS. WH
Suggesting queries to
support task-based search
• We propose an end-to-end generative
probabilistic model
107
108. • Components:
• Source importance
• Document importance
• Keyphrase relevance
q0
Keyphrases
API SUGGS. WEB SNIPPETS WEB DOCS. WH
Suggesting queries to
support task-based search
• We propose an end-to-end generative
probabilistic model
108
109. • Components:
• Source importance
• Document importance
• Keyphrase relevance
• Query suggestion
• We propose an end-to-end generative
probabilistic model
Query suggestions
q0
Keyphrases
API SUGGS. WEB SNIPPETS WEB DOCS. WH
Suggesting queries to
support task-based search
109
110. • We make use of the 2015 and 2016 TREC
Tasks track datasets for the task understanding
problem
• We conduct a principled estimation of the
components, and analyze the best performing
estimators per component
Suggesting queries to
support task-based search
110
111. Suggesting queries to
support task-based search
111
• We observe a heavy reliance on query
suggestions from suggestion APIs
113. • How to jointly generate query suggestions in
query completion and refinement modes?
- Can we do it without relying on log data / API?
• We consider a two-step pipeline:
- Candidate generation
- Candidate ranking
• And focus on the first component
Generating suggestion
candidates
113
114. • We study alternative generation methods and
information sources
- Methods: popular suffix, neural language, sequence-
to-sequence
- Sources: AOL query log, KnowHow, WikiAnswers
• We build a test collection of query suggestion
candidates
Generating suggestion
candidates
114
115. • End-to-end is still the best method overall, but
limited as it depends on API suggestions
• Log data is the most useful information source,
but the other sources provide valuable
suggestions too
• Different method-source configurations
contribute unique suggestions in both modes
Generating suggestion
candidates
115
116. We propose and evaluate a generative
probabilistic model for task-based query
suggestions.
We further study alternative methods and
information sources for suggestion candidate
generation, and build a test collection.
SUMMARY
116
119. Task Recommendation
• The underlying search goal is often a complex
and knowledge-intensive task
• We propose to recommend specific tasks to
users, based on their search queries
119
120. An example
• Planning a wedding reception
wedding reception
Plan a wedding reception
Recommended Tasks:
Plan your wedding reception exit
Announce the bridal party at a reception
Throw a Hawaiian wedding reception
Choose wedding reception activities
120
121. How can we recommend tasks
based on search queries and missions?
121
122. Task Recommendation
• Some terminology:
- Task repository: a catalog of task
descriptions
- Task description: a semi-structured
document that explains the steps involved in
how to complete a given task
- Search mission: a set of queries that all
share the same underlying task
122
123. Task Recommendation
• We introduce two problems:
1. Query-based task recommendation
Given a query, to return a ranked list of tasks
that correspond to the task behind the query
2. Mission-based task recommendation
Given a search mission, to return a ranked list
of recommended tasks, corresponding to the
queries in the mission
123
124. Task Recommendation
• We use a
collection of
WikiHow
articles as
our task
repository
124
How to Make a Wedding Cake
Co-authored by wikiHow Staff ✔
You can make a wedding cake for a customer if you bake for a living,
or you might make a cake for loved one’s wedding to help them save
money. If you love to bake, then you might even want to make your
own wedding cake!
Steps
1 Decide on the number and shape of the cake’s layers. Consider
how many layers and what shape you want the cake to have.
2 Preheat the oven to the temperature indicated by your recipe.
Many recipes call for the oven to be pre-heated to 350 °F (177 °C).
3 Prepare the cake batter according to your recipe’s instructions.
Choose a recipe to create the cake batter for your cake.
4 Pour the batter into a greased, parchment-lined cake pan. Spray
your cake pan with non-stick cooking spray.
Explanation
Main Act
Detailed Act
Title
125. Task Recommendation
• We focus on a subset of tasks, the procedural
tasks
• Procedural task: a search task that can be
accomplished by following a sequence of
specific actions or subtasks
125
126. Task Recommendation
• From a corpus of search queries and missions,
we obtain a set of procedural search missions
126
128. Query-based task
recommendation
• We propose a Learning-to-Rank method for
query-based task recommendation, that
combines a text-based ranking technique with
continuous semantic representations
• We experiment with different word embeddings
and word function sets according to POS-tag
128
131. • To address mission-based task
recommendation, we propose methods that
aggregate the individual query-based
recommendations for each query into mission-
level recommended tasks
131
Mission-based task
recommendation
133. We introduce the problems of query-based and
mission-based task recommendation.
We develop a test collection for task
recommendation, and propose and evaluate
approaches for these problems.
SUMMARY
133
139. Future Directions
139
wedding venue
Det Stavangerske
Klubselskab 120,000 NOK
RESERVE
Olavskleivå 26
Time:
Date:
19:00
Saturday June 27, 2020
Number of
guests:
50-100
Outdoors?
Number of
cars:
Up to 20
Parking?✔
Rosenkildehuset AS
Strandkaien 6
Strømvik allotments
Strømvikveien 1
120,000 NOK
RESERVE
105,000 NOK
RESERVE
Input parameters in
service intents
140. Future Directions
Semantics-aware
query suggestions
Mission-based task
recommendation
140
invitation cards
Make
Homemade
Wedding Cards
Print Your Own
Wedding Invitations
Include a Dress
Code on a
Wedding
Invitation
Queries suggested for invitation cards
invitation card online
invitation card maker
free invitation cards for whatsapp
see-through invitation card
create invitation card with photo free
Recommended tasks