Contextual Search using Cognitive Discovery Capabilities

Similar documents
An Oracle White Paper October Oracle Social Cloud Platform Text Analytics

C. The system is equally reliable for classifying any one of the eight logo types 78% of the time.

Create Swift mobile apps with IBM Watson services IBM Corporation

USER GUIDE DASHBOARD OVERVIEW A STEP BY STEP GUIDE

Information Retrieval

Sharp Social. Natural Language Understanding

Enhancing applications with Cognitive APIs IBM Corporation

Chapter 27 Introduction to Information Retrieval and Web Search

Build a Company List

DMI Exam PDDM Professional Diploma in Digital Marketing Version: 7.0 [ Total Questions: 199 ]

UNIT-V WEB MINING. 3/18/2012 Prof. Asha Ambhaikar, RCET Bhilai.

Maximizing the Value of STM Content through Semantic Enrichment. Frank Stumpf December 1, 2009

IBM Netcool Operations Insight Version 1 Release 4. Integration Guide IBM SC

How Primo Works VE. 1.1 Welcome. Notes: Published by Articulate Storyline Welcome to how Primo works.

Question No : 1 Web spiders carry out a key function within search. What is it? Choose one of the following:

Visualization and text mining of patent and non-patent data

TISA Methodology Threat Intelligence Scoring and Analysis

Empowering People with Knowledge the Next Frontier for Web Search. Wei-Ying Ma Assistant Managing Director Microsoft Research Asia

Search Engine Architecture II

RESEARCH ANALYTICS From Web of Science to InCites. September 20 th, 2010 Marta Plebani

Welcome to Analytics. Welcome to Applause! Table of Contents:

Core Technology Development Team Meeting

Welcome to Your Data

IBM Watson Application Developer Workshop. Watson Knowledge Studio: Building a Machine-learning Annotator with Watson Knowledge Studio.

Embedding Intelligence through Cognitive Services

<is web> Information Systems & Semantic Web University of Koblenz Landau, Germany

DATA MINING - 1DL105, 1DL111

ThreatConnect Learning Exercises

Scuole di dottorato in Bioscienze e biotecnologie e Scienze biomediche sperimentali WEB OF SCIENCE

Google Tools and your Library - the Possibilities are Exponential

Competitive Intelligence and Web Mining:

Parmenides. Semi-automatic. Ontology. construction and maintenance. Ontology. Document convertor/basic processing. Linguistic. Background knowledge

COMPARISON WHITEPAPER. Snowplow Insights VS SaaS load-your-data warehouse providers. We do data collection right.

Learning Ontology-Based User Profiles: A Semantic Approach to Personalized Web Search

TERM BASED WEIGHT MEASURE FOR INFORMATION FILTERING IN SEARCH ENGINES

Ugly PBNer Standard Operating Procedures

Oracle Endeca Information Discovery

TDWI Data Modeling. Data Analysis and Design for BI and Data Warehousing Systems

Chrome based Keyword Visualizer (under sparse text constraint) SANGHO SUH MOONSHIK KANG HOONHEE CHO

Prototyping Data Intensive Apps: TrendingTopics.org

Reputation Management Guide

Finding Stories in Data

Analytics and Visualization

Media Mining Client. Quick User Guide. Version

Front-End Web Developer Nanodegree Syllabus

Oliver Engels & Tillmann Eitelberg. Big Data! Big Quality?

DATA MINING II - 1DL460. Spring 2014"

Troubleshoot DNA Center Using Data Platform

IBM Advantage: IBM Watson Compare and Comply Element Classification

Chapter 2. Architecture of a Search Engine

Overview of Web Mining Techniques and its Application towards Web

Package rzeit2. January 7, 2019

white paper 4 Steps to Better Keyword Grouping Strategies for More Effective & Profitable Keyword Segmentation

Data Mining Concepts & Tasks

Evaluating the Usefulness of Sentiment Information for Focused Crawlers

Data Mining Concepts & Tasks

Blurring the Line Between Developer and Data Scientist

Ranked Retrieval. Evaluation in IR. One option is to average the precision scores at discrete. points on the ROC curve But which points?

Data Analyst Nanodegree Syllabus

THE URBAN COWGIRL PRESENTS KEYWORD RESEARCH

Relevance Feature Discovery for Text Mining

Metadata Standards & Applications. 7. Approaches to Models of Metadata Creation, Storage, and Retrieval

Microsoft FAST Search Server 2010 for SharePoint for Application Developers Course 10806A; 3 Days, Instructor-led

Taming Text. How to Find, Organize, and Manipulate It MANNING GRANT S. INGERSOLL THOMAS S. MORTON ANDREW L. KARRIS. Shelter Island

From Digitally Disrupted to Digital Disrupter. Alex Andrenacci Managing Director Accenture Technology

This is an author-deposited version published in : Eprints ID : 12964

Introduction p. 1 What is the World Wide Web? p. 1 A Brief History of the Web and the Internet p. 2 Web Data Mining p. 4 What is Data Mining? p.

A Survey Of Different Text Mining Techniques Varsha C. Pande 1 and Dr. A.S. Khandelwal 2

Datameer for Data Preparation:

DIGITAL MARKETING Your revolution starts here

Automated Online News Classification with Personalization

PART I A Technical Guide to Oracle Endeca Information Discovery

Natural Language Processing with PoolParty

seobility First steps towards successful SEO

Introduction to Text Mining. Hongning Wang

User guide for GEM-TREND

Flightplan: Getting from Enterprise Search to Cognitive Intelligence

Content Enrichment. An essential strategic capability for every publisher. Enriched content. Delivered.

Executive Alerts User Guide. A walkthrough of all the main features

DEC Computer Technology LESSON 6: DATABASES AND WEB SEARCH ENGINES

Information Retrieval CS Lecture 01. Razvan C. Bunescu School of Electrical Engineering and Computer Science

RPI INSIDE DEEPQA INTRODUCTION QUESTION ANALYSIS 11/26/2013. Watson is. IBM Watson. Inside Watson RPI WATSON RPI WATSON ??? ??? ???

How To Construct A Keyword Strategy?

Google My Business The Free Listing

Mining Web Data. Lijun Zhang

Part I: Data Mining Foundations

6 TOOLS FOR A COMPLETE MARKETING WORKFLOW

USER GUIDE DESIGN A STEP BY STEP GUIDE

SAS Event Stream Processing

On-Page SEO is the foundation with which backlinks and other off-page SEO strategies reach their highest potential.

Ajloun National University

Make the most of your access to ScienceDirect

Enterprise Data Catalog for Microsoft Azure Tutorial

ANALYTICS DATA To Make Better Content Marketing Decisions

Big Data Computing for GIS Data Discovery

Social Business Intelligence in Action

SOURCERER: MINING AND SEARCHING INTERNET- SCALE SOFTWARE REPOSITORIES

The Evolution of Search:

Microsoft FAST Search Server 2010 for SharePoint Evaluation Guide

Text Mining. Representation of Text Documents

Transcription:

Contextual Search using Cognitive Discovery Capabilities In this exercise, you will work with a sample application that uses the Watson Discovery service API s for cognitive search use cases. Discovery service queries are used to extract and detect concepts, keywords, sentiment, entities such as people and companies, relationships as well as trends. Contextual search is a powerful way to gather personalized results based on information from multiple structured and unstructured data repositories. Context is used to help refine and target search results and relevance. Without context, users often must sift through a bunch of irrelevant results before finding what they want. Contextual intelligence helps to assign confidence rankings to search results and streamline the process of finding relevant, current data. The Discovery Service enables developers to build an automated data pipeline to ingest your unstructured data, where the discovery service uses Natural Language Understanding and other cognitive services to enrich understanding of the data. This process consists of automatically tagging NLP meta data, cleansing and normalizing for improved data quality. Once ingested and enriched, queries can be performed. In this exercise, we ll focus on using Discovery service queries with examples for contextual search with news datasources. Exercise: 1. Go to the Discovery News application in Bluemix at: https://discovery-news-demo.mybluemix.net/

Enter a company name for a search term, for instance Amazon. The app uses Discovery service APIs and queries to perform a contextual search in news articles previously ingested and enriched thru the data pipeline and NLP process. It returns the following information: - most frequently occurring topics (concepts), companies, and people (entities) - the news articles search results with links to each article content, along document content sentiment score - positive/negative sentiment rating of the company for content from 10 randomly selected news sources/sites - sentiment trend along a timeline based on mentions of the company along with other companies that it is most frequently mentioned with, in the source articles Examine the contextual search results in each of the four sections: - Top Entities - Top Stories - Sentiment Analysis - Co-mentions & Trends

2. To get ready to examine and understand the Discovery service query that was used to perform this contextual search, lets review a few key concepts: Examples of keyword and entity queries: Queries can be structured for additional options including concepts, sentiment, filtering and aggregations that can provide deeper insights and identify patterns, clusters and trends. The Discovery service provides a query tool that uses a simple query language for multiple query types including boolean, filter, and aggregation queries to discover patterns, trends, and answers.

Aggregations are collections of occurrences of the keywords, concepts and entities from the search results. Aggregations can be nested to extract information and get insights about other keywords, concepts and entities that may be connected or related in the source content. 3. Examine the Discovery service query that was used to perform the news articles contextual search: - In the Top Entities block, click the View Query button. The content of the query consists of the actions and results described in-line in italics below: "return": "title,enrichedtitle.text,url,host,blekko.chrondate", "query": "\"amazon\",language:english", This is a simple keyword query for the company name

"aggregations": [ "nested(enrichedtitle.entities).filter(enrichedtitle.entities.ty pe:company).term(enrichedtitle.entities.text)", This aggregation collects the enriched data for companies the query specifies selection of entities of type company, to get company names mentioned in the news articles "nested(enrichedtitle.entities).filter(enrichedtitle.entities.ty pe:person).term(enrichedtitle.entities.text)", This aggregation collects the enriched data entities of type people to get names of people mentioned in the news articles "term(enrichedtitle.concepts.text)", This aggregation collects the enriched data for concepts, to get the names of topics mentioned in the news articles 4. Click on Response to view the response data returned from the query The response data consists of most frequently occurring company names, names of people and topics, with number of occurrences, in sorted order.

Click on the GoBack button. 5. In the Top Stories section, click on the View Query button, notice that the same query is used to retrieve data for the most frequently appearing stories based on the enriched title and extracted concepts. Click on the Response Data button response data includes the document title, URL, enriched title, host website and sentiment score. Note that in some cases the enriched title may be different than the document title, either to add more context or to remove irrelevant information such as URL strings in titles. 6. In the Sentiment Analysis section, click on the View Query button. Examine the query and notice the stanza in the query that extracts content sentiment: "term(blekko.basedomain).term(docsentiment.type)", This aggregation collects the enriched data for content sentiment of the news articles This next aggregation collects the enriched data for content sentiment of the news articles for min/max sentiment trend along a timeline of each mention of the company plus co-mentioned companies: "term(docsentiment.type)", "min(docsentiment.score)", "max(docsentiment.score)", "filter(enrichedtitle.entities.type::company).term(enrichedtitle.entities.text).timeslice(blekko.chrondate,1day).term(docsentime nt.type)" ], "filter": "blekko.hostrank>20,blekko.chrondate>1491105600,blekko.chrondate <1496376000" Click on the Response Data button response data includes two sections: The first section provides the count of all documents queried having a positive sentiment score, negative sentiment score and neutral sentiment score.

The second section provides the positive/negative/neutral sentiment document count for each of the 10 randomly selected news sites the content was obtained from.

7. In the Co-mentions & Trends section, click on the Response Data button response data includes two sections: The first section provides the count of all documents in which the company name occurs having a positive sentiment score, negative sentiment score and neutral sentiment score, along with individual document sentiment scoring detail data for each. The second section provides for each of the top co-mentioned companies in the documents, the number of matches and the sentiment score document counts of all the documents with the co-mentioned company followed by the individual document sentiment scoring detail data. "key": "Google", "matching_results": 3599, "aggregations": [ "type": "timeslice", "field": "blekko.chrondate", "interval": "1d", "results": [ "key_as_string": "1491091200", "key": 1491091200000, "matching_results": 8, "aggregations": [ "type": "term", "field": "docsentiment.type", "results": [ "key": "negative", "matching_results": 6 }, "key": "positive", "matching_results": 2 }

Learn more about Watson Discovery Service View these education modules on the Watson Discovery service to learn more: https://youtu.be/9ks-ceg6kps https://www.youtube.com/watch/?v=fikhwoj6_fe https://www.youtube.com/watch?v=fikhwoj6_fe&list=plcatfa7iclaycm_ckan7lao- CHG8YUvWx http://event.on24.com/wcc/r/1325563/2f50d5a7f40fb7fa3c1cf0495c9dacbc?partnerref=b WWLP Watson Discovery Service key use cases: Additional use cases are described for financial research, supply chain, customer behavior insights, field engineer advisor and surgical knowledgebase in the Architecture Center for Cognitive Discovery, which also provides detailed information on how to work with the Watson Discovery service to create your document store, create queries, and implement or integrate in your application. Using the Watson Discovery Service getting started documentation and query guide Step by step tutorial on using the Watson Discovery service in Bluemix and building custom queries with the Discovery query tool More information regarding contextual search using cognitive discovery capabilities Complete source code for the application used in this exercise is available at GitHub in the project repository