Machine Learning in Google Algorithm - Where? What? How?
?Download as PPTX, PDF?
1 like?307 views
This talk took place during Engineering and Tech Exhibition organised by University of Bristol on 9th October 2019. In this talk, Rad Paluszak from SUSO Digital will show you how one of the biggest and the most sophisticated companies in the world (Google) is using machine learning in their flagship product - Search. The talk will give you some background to how Google's algorithm works from a web developer and SEO perspective and what elements of it are fluently moving towards being driven by machine learning and neural matching. You will also learn how Google dynamically creates training data sets at scale and ensures they're reliable.
![Deep Learning
Deep learning is part of a broader family of machine learning
methods based on learning data representations (e.g. ontology), as
opposed to task-specific algorithms.
A deep neural network (DNN) is an artificial neural network (ANN)
with multiple layers between the input and output layers. The DNN
finds the correct mathematical manipulation to turn the input into the
output, whether it be a linear relationship or a non-linear
relationship. The network moves through the layers calculating the
probability of each output. ByGlosser.ca[CCBY-SA3.0 (https://creativecommons.org/licenses/by-sa/3.0)],fromWikimediaCommons
https://commons.wikimedia.org/wiki/File:Colored_neural_network.svg](https://image.slidesharecdn.com/radpaluszak-machinelearningingooglealgorithm-191121214143/85/Machine-Learning-in-Google-Algorithm-Where-What-How-6-320.jpg)
![Key Value
First Name [Rad, Radek, Radoslaw]
Last Name [Paluszak]
Gender [M]
Profession [Google, SEO, CTO, Programming,
Web, Webdev, …]
Company [TSI, SUSO Digital, Search
Logistics, The Search Initiative …]
Events [Engineering and Tech Exhibition,
Chiang Mai SEO, Affilliate World,
Marketing Insights]
Universities
Associated
[Poznan University of Technology,
Warwick Business School,
University of Bristol]
Related People [Matt Diggity, Craig Campbell,
Matthew Woodward, …]
… …
Key Value
Name [Engineering and Tech Exhibition]
Organiser [University of Bristol, Lucy
Browning]
Address [Colston Hall, Colston St, Bristol,
BS1 5AR]
Talk Title [Web, Machine Learning, Google,
SEO, Marketing, Algorithms,
Programming]
Related Topics [SEO Ecommerce, Linkbuilding]
Related Entity [SUSO Digital, Google, University
of Bristol]
Coordinates [51° 27' 21.096‘’ N,
2° 35' 51.18‘’ W]
… …](https://image.slidesharecdn.com/radpaluszak-machinelearningingooglealgorithm-191121214143/85/Machine-Learning-in-Google-Algorithm-Where-What-How-20-320.jpg)
![Key Value
First Name [Rad, Radek, Radoslaw]
Last Name [Paluszak]
Gender [M]
Work-related
Entities
[Google, SEO, CTO,
Programming, Web, Webdev, …]
Company [TSI, SUSO Digital, Search
Logistics, The Search Initiative …]
Events [Engineering and Tech
Exhibition, Chiang Mai SEO,
Affilliate World, Marketing
Insights]
Universities
Associated
[Poznan University of Technology,
Warwick Business School,
University of Bristol]
Related People [Matt Diggity, Craig Campbell,
Matthew Woodward, …]
… …
Key Value
Name [Engineering and Tech
Exhibition]
Organiser [University of Bristol, Lucy
Browning]
Address [Colston Hall, Colston St, Bristol,
BS1 5AR]
Talk Topics [Web, Machine Learning,
Google, SEO, Marketing,
Algorithms, Programming]
Related Topics [SEO Ecommerce, Linkbuilding]
Related Entity [SUSO Digital, Google, University
of Bristol]
Coordinates [51° 27' 21.096‘’ N,
2° 35' 51.18‘’ W]
… …](https://image.slidesharecdn.com/radpaluszak-machinelearningingooglealgorithm-191121214143/85/Machine-Learning-in-Google-Algorithm-Where-What-How-21-320.jpg)
Machine Learning in Google Algorithm - Where? What? How?
- 1. Machine Learning in Google Algorithm Where? What? How? Rad Paluszak
- 2. Machine Learning in Google Algorithm ? Machine Learning ? Deep Learning ? Hummingbird ? RankBrain ? Natural Language Processing & Understanding ? Neural Matching ? Crawling and Indexing ? Penalty Process ? Large Scale Training Data Sets ? Summary What am I talking about? Machine Learning in Google Algorithm: Where? What? How?
- 3. Rad Paluszak ? “SEO” birthday - 2010 (Caffeine update) ? Web developer “at heart” ? Algorithms <3 ? Machine Learning <3 ? Data Mining <3 ? “Technical SEO Artist” @radpaluszak ? rad@paluszak.me ? rad@susodigital.com Director of Technology
- 4. Machine Learning Machine learning is a method of data analysis that automates analytical model building. It is a branch of artificial intelligence based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.
- 5. Machine Learning Machine learning is a technology which, instead of programming the computers in a very precise manner, allows them to perform tasks based on what they learn through data analysis.
- 6. Deep Learning Deep learning is part of a broader family of machine learning methods based on learning data representations (e.g. ontology), as opposed to task-specific algorithms. A deep neural network (DNN) is an artificial neural network (ANN) with multiple layers between the input and output layers. The DNN finds the correct mathematical manipulation to turn the input into the output, whether it be a linear relationship or a non-linear relationship. The network moves through the layers calculating the probability of each output. ByGlosser.ca[CCBY-SA3.0 (https://creativecommons.org/licenses/by-sa/3.0)],fromWikimediaCommons https://commons.wikimedia.org/wiki/File:Colored_neural_network.svg
- 7. Hummingbird Codename given to a significant algorithm change in Google Search in 2013. Its name was derived from the speed and accuracy of the hummingbird Conversational Search Natural Language Processing Query Intent Semantic Model Analysis New Ranking Signals Importance of Authority Long-tail Focused
- 17. Google Search Results ? Over 70k searches every SECOND ? 15% of queries have never been seen by Google ? ~500 million brand new queries a day https://www.cnet.com/news/google-search-scratches-its-brain-500-million-times-a-day/
- 18. RankBrain RankBrain is an artificial intelligence system, the use of which was confirmed by Google on 26 October 2015. It helps Google to process search results and provide more relevant search results for users. If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the results, accordingly, making it more effective at handling never-before-seen search queries or keywords.
- 19. RankBrain Understands similarity of the queries based on multi- dimensional vector space analysis & the proximity of one query to the other.
- 20. Key Value First Name [Rad, Radek, Radoslaw] Last Name [Paluszak] Gender [M] Profession [Google, SEO, CTO, Programming, Web, Webdev, …] Company [TSI, SUSO Digital, Search Logistics, The Search Initiative …] Events [Engineering and Tech Exhibition, Chiang Mai SEO, Affilliate World, Marketing Insights] Universities Associated [Poznan University of Technology, Warwick Business School, University of Bristol] Related People [Matt Diggity, Craig Campbell, Matthew Woodward, …] … … Key Value Name [Engineering and Tech Exhibition] Organiser [University of Bristol, Lucy Browning] Address [Colston Hall, Colston St, Bristol, BS1 5AR] Talk Title [Web, Machine Learning, Google, SEO, Marketing, Algorithms, Programming] Related Topics [SEO Ecommerce, Linkbuilding] Related Entity [SUSO Digital, Google, University of Bristol] Coordinates [51° 27' 21.096‘’ N, 2° 35' 51.18‘’ W] … …
- 21. Key Value First Name [Rad, Radek, Radoslaw] Last Name [Paluszak] Gender [M] Work-related Entities [Google, SEO, CTO, Programming, Web, Webdev, …] Company [TSI, SUSO Digital, Search Logistics, The Search Initiative …] Events [Engineering and Tech Exhibition, Chiang Mai SEO, Affilliate World, Marketing Insights] Universities Associated [Poznan University of Technology, Warwick Business School, University of Bristol] Related People [Matt Diggity, Craig Campbell, Matthew Woodward, …] … … Key Value Name [Engineering and Tech Exhibition] Organiser [University of Bristol, Lucy Browning] Address [Colston Hall, Colston St, Bristol, BS1 5AR] Talk Topics [Web, Machine Learning, Google, SEO, Marketing, Algorithms, Programming] Related Topics [SEO Ecommerce, Linkbuilding] Related Entity [SUSO Digital, Google, University of Bristol] Coordinates [51° 27' 21.096‘’ N, 2° 35' 51.18‘’ W] … …
- 22. Rank Brain in Search Results User Query Results Satisfaction Analysis (CTR, BR) Results Post process ing Relevance Matching Intent AnalysisNLPQuery Parsing
- 23. Rank Brain in Search Results
- 26. Rank Brain in Search Results Query Matching Purpose Instant Answer Latent Intent Interpretation Semantic Structure
- 27. Natural Language Processing CCBY-SA3.0, https://en.wikipedia.org/w/index.php?curid=11556338 Apache OpenNLP ? S Simple declarative phrase ? NP Noun phrase ? VP Verb phrase ? DT Determiner ? JJ Adjective ? NN Noun, singular or mass ? VBZ Verb, 3rd person singular present ? VBN Verb, past participle ? TO to http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.9.8216&rep=rep1&type=pdf
- 29. Neural Matching Neural Matching – AI method to better connect words to concepts. Introduced in 2018 (officially confirmed in September 2018). Affects ~30% of all queries.
- 31. “Google always tries to predict your site structure and assess what’s worth crawling and what is not.”
- 34. I didn’t learn anything new about the site structure. GooglebotImageCredits: http://www.thesempost.com/blocking-googlebot-with-bad-bot-scripts-wordfence/ https://www.seroundtable.com/google-crawl-report-problem-19894.html Schedulerforsearchenginecrawler https://patents.google.com/patent/US7725452B1/en
- 35. “Manual Penalty process starts with a classification of suspicious patterns detected by machine learning algorithms.”
- 36. Manual Penalty Manual Verifi- cation* Classifi- cation Pre- Screen- ing Machine Learning Machine Learning in Link Assessment
- 37. Machine Learning in Link Assessment
- 38. Training Data Set – Search Quality Evaluators https://static.googleusercontent.com/media/guidelines.raterhub.com/en//searchqualityevaluatorguidelines.pdf
- 39. Machine Learning in Google ? Understanding Query ? Intent (Understanding & Prediction) ? Result Matching (Core Algo) ? First-Time-Query Matching ? Neural Matching ? Algorithmic Devaluation ? Algorithmic Quality Factors Analysis ? Resource Management (Crawl & Index) ? “Manual” Action Process (Penalty) ? Image Recognition ? Fake News Detection ? Ethical Filters ? Location Prediction Google Search Other Products ? Translation(Google Translate) ? Ad Placement & Interest Matching (Google Ads) ? Security (all products, Safe Browsing) ? Recommendations(YouTube) ? Travel (Google Maps) ? Reviews (Google My Business) ? Surveys (Google Rewards) ? Text Generation (Gmail) ? Behaviour Patterns (Google Home, Google Chrome, Nest) ? Recommendations (Google Now, Discover) ? Tensor Flow (Google Cloud)
- 41. Thank You! RAD PALUSZAK Director of Technology at SUSO Digital rad@susodigital.com rad@paluszak.me @radpaluszak Email Subject: Bristol Machine Learning Talk 2019