SlideShare a Scribd company logo
DataScience@SMU
Language Empowered Recommendation
Albert Asuncion, Peter Kouvaris, Ekaterina Pirogova, Hari Sanadhya, Arun Rajagopal
Master of Science in Data Science
Southern Methodist University, Dallas, TX 75275, USA
Aligning the Stars
Our star factor leverages two convolutional neural networks
that are tasked with learning how review text affects ratings
for users and businesses separately.
These results are combined to produce a new star rating
based on a mixture of the user, item, and review text.
An example of this would be if User A frequently wrote
negatively worded reviews, but upon going to Business B
wrote a highly positive one, our first neural network would
weight this derivation very highly.
Our other network would consider the same logic for
Business B as if it was a user giving reviews above. If all
user’s generally give Business B positive reviews, this should
be considered and produce a slightly lower star rating. The
average of these is used.
Ethics
Measuring Success
The success of a star rating system can be measured by the
normalization of data and the improvement in
recommendations on a subset of sample data. Our method
performs 22% better using the FCP statistic.
The importance of a recommender's systems results on
users and businesses can be drastic, requiring there to be
clear laws and privacy rights associated with these. Think of
these examples:
• An influential Yelp user is threatened by a business.
• A small business is banned from Yelp.
• Lack of privacy for users
Recommender systems are ubiquitous, driving us to web and
mobile applications for help deciding which movies to see,
books of interest to read, or restaurants to enjoy our next great
meal. We examined one of the most widely known and
recognized of these recommender systems and improved
upon its star rating and recommender system. More
specifically, we examined the following questions:
• How do we make ordinal star ratings more meaningful?
• How can we improve Yelp recommendations?
Introduction
Conclusions
• The use of Natural Language Processing makes star
ratings more meaningful.
• The text of review greatly improves the recommendation
system quality.
• Fake reviews penalty is too harsh in Yelp system.
1st Gen: Item, User, Rating
2nd Gen: Items, Users, Rating, Context, Time, Location,
Rating
3rd Gen: Ontologies (Complex informational web structures)
Evolution of Recommenders
The Yelp star rating is ambiguous and fraught with issues:
1. Yelp data is inherently handicapped
a) User time of visit is unknown
b) Most on non-mobile app users are not registered
c) Recommendation success/failure is difficult to measure
2. Ratings are biased (see Figure 1)
3. Distances between rating values are not equidistant
Failing Star
Collaborative filtering algorithms rely on information about
other users who have interacted with the item of interest.
These algorithms benefit greatly from numeric ratings that
scale in a linear way. This feature detailed in step 2 of Figure
3, allows us to :
• Provide better recommendations at lower training sizes.
• Leverage star ratings for filtering on other dimensions.
Star ratings are simple to understand and applied to a variety
of use cases. Building useful data metrics like these can
benefit more than just recommender models.
Importance of Star Shape
+ =
Basic Stars Sentiment Stars Deep Learning Stars
Figure 1: Using deep learning and language features we eliminate bias and
create equidistance between ratings.
Figure 2: Star distribution shift, improving from left to right.
Figure 3: Recommender system pipeline
Basic NLP
0.4307 0.5289

More Related Content

What's hot

Language Empowered Recommendations
Language Empowered RecommendationsLanguage Empowered Recommendations
Language Empowered RecommendationsHari Sanadhya
 
Reflection of Statistical Analysis.BISM
Reflection of Statistical Analysis.BISMReflection of Statistical Analysis.BISM
Reflection of Statistical Analysis.BISMAshley Kruempel
 
Thesis Presentation
Thesis PresentationThesis Presentation
Thesis Presentationnirvdrum
 
[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...
[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...
[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...YONG ZHENG
 
Recommendation System for Design Patterns in Software Development
Recommendation System for Design Patterns in Software DevelopmentRecommendation System for Design Patterns in Software Development
Recommendation System for Design Patterns in Software DevelopmentFrancis Palma
 
Fortner_OSCARPresentation
Fortner_OSCARPresentationFortner_OSCARPresentation
Fortner_OSCARPresentationAshley Fortner
 
Movies recommendation system in R Studio, Machine learning
Movies recommendation system in  R Studio, Machine learning Movies recommendation system in  R Studio, Machine learning
Movies recommendation system in R Studio, Machine learning Mauryasuraj98
 
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User StudiesYONG ZHENG
 
Multiple Response Questions - Allowing for chance in authentic assessments
Multiple Response Questions - Allowing for chance in authentic assessmentsMultiple Response Questions - Allowing for chance in authentic assessments
Multiple Response Questions - Allowing for chance in authentic assessmentsMhairi Mcalpine
 
Place aware content selection
Place aware content selectionPlace aware content selection
Place aware content selectionJehangir Khan
 
Machine learning meets user analytics - Metageni tech talk
Machine learning meets user analytics - Metageni tech talkMachine learning meets user analytics - Metageni tech talk
Machine learning meets user analytics - Metageni tech talkGabriel Hughes PhD
 
PosterPresentations-36x48-final_final_final_publish_1
PosterPresentations-36x48-final_final_final_publish_1PosterPresentations-36x48-final_final_final_publish_1
PosterPresentations-36x48-final_final_final_publish_1Mark Davenport
 
Project Data Incorporating Qualitative Factors for Improved Software Defect P...
Project Data Incorporating Qualitative Factors for Improved Software Defect P...Project Data Incorporating Qualitative Factors for Improved Software Defect P...
Project Data Incorporating Qualitative Factors for Improved Software Defect P...Tim Menzies
 
AIRG Presentation
AIRG PresentationAIRG Presentation
AIRG Presentationnirvdrum
 
organization-public relationships
organization-public relationshipsorganization-public relationships
organization-public relationshipsAlyssa Gonzalez
 
Crowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral OutcomesCrowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral OutcomesAlekya Yermal
 
SENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTION
SENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTIONSENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTION
SENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTIONIncedo
 

What's hot (20)

Language Empowered Recommendations
Language Empowered RecommendationsLanguage Empowered Recommendations
Language Empowered Recommendations
 
Reflection of Statistical Analysis.BISM
Reflection of Statistical Analysis.BISMReflection of Statistical Analysis.BISM
Reflection of Statistical Analysis.BISM
 
Recommender Systems
Recommender SystemsRecommender Systems
Recommender Systems
 
Thesis Presentation
Thesis PresentationThesis Presentation
Thesis Presentation
 
[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...
[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...
[ADMA 2017] Identification of Grey Sheep Users By Histogram Intersection In R...
 
Recommendation System for Design Patterns in Software Development
Recommendation System for Design Patterns in Software DevelopmentRecommendation System for Design Patterns in Software Development
Recommendation System for Design Patterns in Software Development
 
Fortner_OSCARPresentation
Fortner_OSCARPresentationFortner_OSCARPresentation
Fortner_OSCARPresentation
 
Movies recommendation system in R Studio, Machine learning
Movies recommendation system in  R Studio, Machine learning Movies recommendation system in  R Studio, Machine learning
Movies recommendation system in R Studio, Machine learning
 
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
[WI 2017] Context Suggestion: Empirical Evaluations vs User Studies
 
Multiple Response Questions - Allowing for chance in authentic assessments
Multiple Response Questions - Allowing for chance in authentic assessmentsMultiple Response Questions - Allowing for chance in authentic assessments
Multiple Response Questions - Allowing for chance in authentic assessments
 
Place aware content selection
Place aware content selectionPlace aware content selection
Place aware content selection
 
Machine learning meets user analytics - Metageni tech talk
Machine learning meets user analytics - Metageni tech talkMachine learning meets user analytics - Metageni tech talk
Machine learning meets user analytics - Metageni tech talk
 
PosterPresentations-36x48-final_final_final_publish_1
PosterPresentations-36x48-final_final_final_publish_1PosterPresentations-36x48-final_final_final_publish_1
PosterPresentations-36x48-final_final_final_publish_1
 
Project Data Incorporating Qualitative Factors for Improved Software Defect P...
Project Data Incorporating Qualitative Factors for Improved Software Defect P...Project Data Incorporating Qualitative Factors for Improved Software Defect P...
Project Data Incorporating Qualitative Factors for Improved Software Defect P...
 
AIRG Presentation
AIRG PresentationAIRG Presentation
AIRG Presentation
 
organization-public relationships
organization-public relationshipsorganization-public relationships
organization-public relationships
 
Crowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral OutcomesCrowdsourcing Predictors of Behavioral Outcomes
Crowdsourcing Predictors of Behavioral Outcomes
 
Marketing Research
Marketing ResearchMarketing Research
Marketing Research
 
SENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTION
SENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTIONSENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTION
SENTIMENT ANALYSIS FOR DRUG DEVELOPMENT AND PROMOTION
 
Which is More Important in Online Review Usefulness, Heuristic or Systematic ...
Which is More Important in Online Review Usefulness, Heuristic or Systematic ...Which is More Important in Online Review Usefulness, Heuristic or Systematic ...
Which is More Important in Online Review Usefulness, Heuristic or Systematic ...
 

Similar to Text Enhanced Recommendation System Model Based on Yelp Reviews

IRJET- Predicting Review Ratings for Product Marketing
IRJET- Predicting Review Ratings for Product MarketingIRJET- Predicting Review Ratings for Product Marketing
IRJET- Predicting Review Ratings for Product MarketingIRJET Journal
 
Online Service Rating Prediction by Removing Paid Users and Jaccard Coefficient
Online Service Rating Prediction by Removing Paid Users and Jaccard CoefficientOnline Service Rating Prediction by Removing Paid Users and Jaccard Coefficient
Online Service Rating Prediction by Removing Paid Users and Jaccard CoefficientIRJET Journal
 
Automatic Recommendation of Trustworthy Users in Online Product Rating Sites
Automatic Recommendation of Trustworthy Users in Online Product Rating SitesAutomatic Recommendation of Trustworthy Users in Online Product Rating Sites
Automatic Recommendation of Trustworthy Users in Online Product Rating SitesIRJET Journal
 
Performance appraisal evaluation
Performance appraisal evaluationPerformance appraisal evaluation
Performance appraisal evaluationjuniecarter5
 
session2.pdf
session2.pdfsession2.pdf
session2.pdfshero2015
 
Can you trust online ratings a mutual
Can you trust online ratings a mutualCan you trust online ratings a mutual
Can you trust online ratings a mutualAbhishek Kb
 
A Supervised Modeling Approach to Determine Elite Status of Yelp Members
A Supervised Modeling Approach to Determine Elite Status of Yelp MembersA Supervised Modeling Approach to Determine Elite Status of Yelp Members
A Supervised Modeling Approach to Determine Elite Status of Yelp MembersJennifer (Hui) Li
 
Literature review of performance appraisal
Literature review of performance appraisalLiterature review of performance appraisal
Literature review of performance appraisalkeshiaflores440
 
Rating System Algorithms Document
Rating System Algorithms DocumentRating System Algorithms Document
Rating System Algorithms DocumentScandala Tamang
 
IRJET- Survey of Classification of Business Reviews using Sentiment Analysis
IRJET- Survey of Classification of Business Reviews using Sentiment AnalysisIRJET- Survey of Classification of Business Reviews using Sentiment Analysis
IRJET- Survey of Classification of Business Reviews using Sentiment AnalysisIRJET Journal
 
IRJET- Analysis of Question and Answering Recommendation System
IRJET-  	  Analysis of Question and Answering Recommendation SystemIRJET-  	  Analysis of Question and Answering Recommendation System
IRJET- Analysis of Question and Answering Recommendation SystemIRJET Journal
 
Problems in performance appraisal
Problems in performance appraisalProblems in performance appraisal
Problems in performance appraisalmarianelson0292
 
IRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User InterestIRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User InterestIRJET Journal
 
IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...
IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...
IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...IRJET Journal
 
Performance appraisal rating
Performance appraisal ratingPerformance appraisal rating
Performance appraisal ratingalexwalker024
 
IRJET-A Novel Technic to Notice Spam Reviews on e-Shopping
IRJET-A Novel Technic to Notice Spam Reviews on e-ShoppingIRJET-A Novel Technic to Notice Spam Reviews on e-Shopping
IRJET-A Novel Technic to Notice Spam Reviews on e-ShoppingIRJET Journal
 
Performance appraisal phrase
Performance appraisal phrasePerformance appraisal phrase
Performance appraisal phraserogeryoung116
 
Detection of Fraud Reviews for a Product
Detection of Fraud Reviews for a ProductDetection of Fraud Reviews for a Product
Detection of Fraud Reviews for a ProductIJSRD
 
Performance appraisal statistics
Performance appraisal statisticsPerformance appraisal statistics
Performance appraisal statisticsaprileward14
 

Similar to Text Enhanced Recommendation System Model Based on Yelp Reviews (20)

IRJET- Predicting Review Ratings for Product Marketing
IRJET- Predicting Review Ratings for Product MarketingIRJET- Predicting Review Ratings for Product Marketing
IRJET- Predicting Review Ratings for Product Marketing
 
Online Service Rating Prediction by Removing Paid Users and Jaccard Coefficient
Online Service Rating Prediction by Removing Paid Users and Jaccard CoefficientOnline Service Rating Prediction by Removing Paid Users and Jaccard Coefficient
Online Service Rating Prediction by Removing Paid Users and Jaccard Coefficient
 
Automatic Recommendation of Trustworthy Users in Online Product Rating Sites
Automatic Recommendation of Trustworthy Users in Online Product Rating SitesAutomatic Recommendation of Trustworthy Users in Online Product Rating Sites
Automatic Recommendation of Trustworthy Users in Online Product Rating Sites
 
Performance appraisal evaluation
Performance appraisal evaluationPerformance appraisal evaluation
Performance appraisal evaluation
 
session2.pdf
session2.pdfsession2.pdf
session2.pdf
 
Can you trust online ratings a mutual
Can you trust online ratings a mutualCan you trust online ratings a mutual
Can you trust online ratings a mutual
 
A Supervised Modeling Approach to Determine Elite Status of Yelp Members
A Supervised Modeling Approach to Determine Elite Status of Yelp MembersA Supervised Modeling Approach to Determine Elite Status of Yelp Members
A Supervised Modeling Approach to Determine Elite Status of Yelp Members
 
Literature review of performance appraisal
Literature review of performance appraisalLiterature review of performance appraisal
Literature review of performance appraisal
 
Rating System Algorithms Document
Rating System Algorithms DocumentRating System Algorithms Document
Rating System Algorithms Document
 
IRJET- Survey of Classification of Business Reviews using Sentiment Analysis
IRJET- Survey of Classification of Business Reviews using Sentiment AnalysisIRJET- Survey of Classification of Business Reviews using Sentiment Analysis
IRJET- Survey of Classification of Business Reviews using Sentiment Analysis
 
IRJET- Analysis of Question and Answering Recommendation System
IRJET-  	  Analysis of Question and Answering Recommendation SystemIRJET-  	  Analysis of Question and Answering Recommendation System
IRJET- Analysis of Question and Answering Recommendation System
 
B05110409
B05110409B05110409
B05110409
 
Problems in performance appraisal
Problems in performance appraisalProblems in performance appraisal
Problems in performance appraisal
 
IRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User InterestIRJET- Analysis of Rating Difference and User Interest
IRJET- Analysis of Rating Difference and User Interest
 
IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...
IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...
IRJET - Characterizing Products’ Outcome by Sentiment Analysis and Predicting...
 
Performance appraisal rating
Performance appraisal ratingPerformance appraisal rating
Performance appraisal rating
 
IRJET-A Novel Technic to Notice Spam Reviews on e-Shopping
IRJET-A Novel Technic to Notice Spam Reviews on e-ShoppingIRJET-A Novel Technic to Notice Spam Reviews on e-Shopping
IRJET-A Novel Technic to Notice Spam Reviews on e-Shopping
 
Performance appraisal phrase
Performance appraisal phrasePerformance appraisal phrase
Performance appraisal phrase
 
Detection of Fraud Reviews for a Product
Detection of Fraud Reviews for a ProductDetection of Fraud Reviews for a Product
Detection of Fraud Reviews for a Product
 
Performance appraisal statistics
Performance appraisal statisticsPerformance appraisal statistics
Performance appraisal statistics
 

Recently uploaded

一比一原版(TWU毕业证)西三一大学毕业证成绩单
一比一原版(TWU毕业证)西三一大学毕业证成绩单一比一原版(TWU毕业证)西三一大学毕业证成绩单
一比一原版(TWU毕业证)西三一大学毕业证成绩单ocavb
 
一比一原版(BU毕业证)波士顿大学毕业证成绩单
一比一原版(BU毕业证)波士顿大学毕业证成绩单一比一原版(BU毕业证)波士顿大学毕业证成绩单
一比一原版(BU毕业证)波士顿大学毕业证成绩单ewymefz
 
tapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive datatapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive datatheahmadsaood
 
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP
 
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...Domenico Conte
 
2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...
2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...
2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...elinavihriala
 
Computer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage sComputer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage sMAQIB18
 
Uber Ride Supply Demand Gap Analysis Report
Uber Ride Supply Demand Gap Analysis ReportUber Ride Supply Demand Gap Analysis Report
Uber Ride Supply Demand Gap Analysis ReportSatyamNeelmani2
 
社内勉強会資料_LLM Agents                              .
社内勉強会資料_LLM Agents                              .社内勉強会資料_LLM Agents                              .
社内勉強会資料_LLM Agents                              .NABLAS株式会社
 
How can I successfully sell my pi coins in Philippines?
How can I successfully sell my pi coins in Philippines?How can I successfully sell my pi coins in Philippines?
How can I successfully sell my pi coins in Philippines?DOT TECH
 
Opendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptxOpendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptxOpendatabay
 
Using PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDBUsing PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDBAlireza Kamrani
 
Supply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflictSupply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflictJack Cole
 
一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单ewymefz
 
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单ewymefz
 
Tabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsTabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsalex933524
 
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单vcaxypu
 
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单nscud
 
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单ewymefz
 

Recently uploaded (20)

一比一原版(TWU毕业证)西三一大学毕业证成绩单
一比一原版(TWU毕业证)西三一大学毕业证成绩单一比一原版(TWU毕业证)西三一大学毕业证成绩单
一比一原版(TWU毕业证)西三一大学毕业证成绩单
 
一比一原版(BU毕业证)波士顿大学毕业证成绩单
一比一原版(BU毕业证)波士顿大学毕业证成绩单一比一原版(BU毕业证)波士顿大学毕业证成绩单
一比一原版(BU毕业证)波士顿大学毕业证成绩单
 
tapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive datatapal brand analysis PPT slide for comptetive data
tapal brand analysis PPT slide for comptetive data
 
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
 
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
Professional Data Engineer Certification Exam Guide  _  Learn  _  Google Clou...
 
2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...
2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...
2024-05-14 - Tableau User Group - TC24 Hot Topics - Tableau Pulse and Einstei...
 
Computer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage sComputer Presentation.pptx ecommerce advantage s
Computer Presentation.pptx ecommerce advantage s
 
Uber Ride Supply Demand Gap Analysis Report
Uber Ride Supply Demand Gap Analysis ReportUber Ride Supply Demand Gap Analysis Report
Uber Ride Supply Demand Gap Analysis Report
 
社内勉強会資料_LLM Agents                              .
社内勉強会資料_LLM Agents                              .社内勉強会資料_LLM Agents                              .
社内勉強会資料_LLM Agents                              .
 
How can I successfully sell my pi coins in Philippines?
How can I successfully sell my pi coins in Philippines?How can I successfully sell my pi coins in Philippines?
How can I successfully sell my pi coins in Philippines?
 
Opendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptxOpendatabay - Open Data Marketplace.pptx
Opendatabay - Open Data Marketplace.pptx
 
Slip-and-fall Injuries: Top Workers' Comp Claims
Slip-and-fall Injuries: Top Workers' Comp ClaimsSlip-and-fall Injuries: Top Workers' Comp Claims
Slip-and-fall Injuries: Top Workers' Comp Claims
 
Using PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDBUsing PDB Relocation to Move a Single PDB to Another Existing CDB
Using PDB Relocation to Move a Single PDB to Another Existing CDB
 
Supply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflictSupply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflict
 
一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单
 
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
 
Tabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsTabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflows
 
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
一比一原版(ArtEZ毕业证)ArtEZ艺术学院毕业证成绩单
 
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
 
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
 

Text Enhanced Recommendation System Model Based on Yelp Reviews

  • 1. DataScience@SMU Language Empowered Recommendation Albert Asuncion, Peter Kouvaris, Ekaterina Pirogova, Hari Sanadhya, Arun Rajagopal Master of Science in Data Science Southern Methodist University, Dallas, TX 75275, USA Aligning the Stars Our star factor leverages two convolutional neural networks that are tasked with learning how review text affects ratings for users and businesses separately. These results are combined to produce a new star rating based on a mixture of the user, item, and review text. An example of this would be if User A frequently wrote negatively worded reviews, but upon going to Business B wrote a highly positive one, our first neural network would weight this derivation very highly. Our other network would consider the same logic for Business B as if it was a user giving reviews above. If all user’s generally give Business B positive reviews, this should be considered and produce a slightly lower star rating. The average of these is used. Ethics Measuring Success The success of a star rating system can be measured by the normalization of data and the improvement in recommendations on a subset of sample data. Our method performs 22% better using the FCP statistic. The importance of a recommender's systems results on users and businesses can be drastic, requiring there to be clear laws and privacy rights associated with these. Think of these examples: • An influential Yelp user is threatened by a business. • A small business is banned from Yelp. • Lack of privacy for users Recommender systems are ubiquitous, driving us to web and mobile applications for help deciding which movies to see, books of interest to read, or restaurants to enjoy our next great meal. We examined one of the most widely known and recognized of these recommender systems and improved upon its star rating and recommender system. More specifically, we examined the following questions: • How do we make ordinal star ratings more meaningful? • How can we improve Yelp recommendations? Introduction Conclusions • The use of Natural Language Processing makes star ratings more meaningful. • The text of review greatly improves the recommendation system quality. • Fake reviews penalty is too harsh in Yelp system. 1st Gen: Item, User, Rating 2nd Gen: Items, Users, Rating, Context, Time, Location, Rating 3rd Gen: Ontologies (Complex informational web structures) Evolution of Recommenders The Yelp star rating is ambiguous and fraught with issues: 1. Yelp data is inherently handicapped a) User time of visit is unknown b) Most on non-mobile app users are not registered c) Recommendation success/failure is difficult to measure 2. Ratings are biased (see Figure 1) 3. Distances between rating values are not equidistant Failing Star Collaborative filtering algorithms rely on information about other users who have interacted with the item of interest. These algorithms benefit greatly from numeric ratings that scale in a linear way. This feature detailed in step 2 of Figure 3, allows us to : • Provide better recommendations at lower training sizes. • Leverage star ratings for filtering on other dimensions. Star ratings are simple to understand and applied to a variety of use cases. Building useful data metrics like these can benefit more than just recommender models. Importance of Star Shape + = Basic Stars Sentiment Stars Deep Learning Stars Figure 1: Using deep learning and language features we eliminate bias and create equidistance between ratings. Figure 2: Star distribution shift, improving from left to right. Figure 3: Recommender system pipeline Basic NLP 0.4307 0.5289