The User is the Query - The Rise of Predictive Proactive Search
38 likes7,725 views
The document discusses the evolution of search technology, emphasizing the shift towards predictive, proactive search where user behavior and preferences are essential. It highlights challenges in understanding intent and the variety of ways users express their queries, along with the increasing prominence of visual information and personalized content. Key developments like Google Discover underscore the importance of adapting to user needs by moving beyond traditional query-based approaches to more context-aware, task-driven systems.
![Let’s Take The Query [Easter]](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-60-320.jpg)
![[Four candles] + [fork handles] interest over time](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-73-320.jpg)
![References
• Broder, A., 2002, September. A taxonomy of web search. In ACM Sigir forum (Vol. 36, No.
2, pp. 3-10). ACM.
• Chuklin, A., Severyn, A., Trippas, J., Alfonseca, E., Silen, H. and Spina, D., 2018. Prosody
Modifications for Question-Answering in Voice-Only Settings. arXiv preprint
arXiv:1806.03957.
• HigherVisibility. 2018. How Popular is Voice Search? | HigherVisibility. [ONLINE] Available
at: https://www.highervisibility.com/blog/how-popular-is-voice-search/
• Filippova, K., Alfonseca, E., Colmenares, C.A., Kaiser, L. and Vinyals, O., 2015. Sentence
compression by deletion with lstms. In Proceedings of the 2015 Conference on Empirical
Methods in Natural Language Processing (pp. 360-368).
• Filippova, K. and Alfonseca, E., 2015. Fast k-best sentence compression. arXiv preprint
arXiv:1510.08418.
• Google Developers. 2018. Content-based Actions | Actions on Google | Google
Developers. [ONLINE] Available at: https://developers.google.com/actions/content-
actions/. [Accessed 18 June 2018]](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-182-320.jpg)
![References
• Mitkov, R., 2014. Anaphora resolution. Routledge.
• NLP Department - Stanford University - Imran Q Sayed. 2018. Issues in Anaphora Resolution.
[ONLINE] Available
at: https://nlp.stanford.edu/courses/cs224n/2003/fp/iqsayed/project_report.pdf. [Accessed 28
June 2018].
• Radlinski, F. and Craswell, N., 2017, March. A theoretical framework for conversational search.
In Proceedings of the 2017 Conference on Conference Human Information Interaction and
Retrieval (pp. 117-126). ACM.
• Schalkwyk, J., Beeferman, D., Beaufays, F., Byrne, B., Chelba, C., Cohen, M., Kamvar, M. and
Strope, B., 2010. “Your word is my command”: Google search by voice: a case study. In Advances
in speech recognition (pp. 61-90). Springer, Boston, MA.
• SISTRIX. 2018. Stepping out of the SEO Bubble - SISTRIX. [ONLINE] Available
at: https://www.sistrix.com/blog/stepping-out-of-the-seo-bubble/. [Accessed 16 June 2018].
• Presentation at ESSIR2017 on work by Radinsky, K., Svore, K.M., Dumais, S.T., Shokouhi, M.,
Teevan, J., Bocharov, A. and Horvitz, E., 2013. Behavioral dynamics on the web: Learning,
modeling, and prediction. ACM Transactions on Information Systems (TOIS), 31(3), p.16.](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-183-320.jpg)
![References
• The Stanford Question Answering Dataset. 2018. The Stanford
Question Answering Dataset. [ONLINE] Available
at: https://rajpurkar.github.io/SQuAD-explorer/.
• Trippas, J.R., Spina, D., Cavedon, L., Joho, H. and Sanderson, M., 2018.
Informing the Design of Spoken Conversational Search.
• https://medium.com/@ashishgupta031/sequence-aware-
reinforcement-learning-over-knowledge-graphs-a8af155e716c
• Jansen, B.J., Booth, D.L. and Spink, A., 2008. Determining the
informational, navigational, and transactional intent of Web
queries. Information Processing & Management, 44(3), pp.1251-
1266.](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-184-320.jpg)
![References
Radinsky, K., Svore, K.M., Dumais, S.T., Shokouhi, M., Teevan, J., Bocharov, A. and Horvitz, E., 2013.
Behavioral dynamics on the web: Learning, modeling, and prediction. ACM Transactions on Information
Systems (TOIS), 31(3), p.16
Sadikov, E., Madhavan, J. and Halevy, A., Google LLC, 2013. Clustering query
refinements by inferred user intent. U.S. Patent 8,423,538.
Official Google Webmaster Central Blog. 2019. Official Google Webmaster Central
Blog: Rolling out mobile-first indexing . [ONLINE] Available
at: https://webmasters.googleblog.com/2018/03/rolling-out-mobile-first-
indexing.html. [Accessed 25 September 2019].
Zhou, S., Cheng, K. and Men, L., 2017, April. The survey of large-scale query
classification. In AIP Conference Proceedings (Vol. 1834, No. 1, p. 040045). AIP
Publishing.](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-185-320.jpg)
![References
Search Engine Land. 2019. Starting July 1, all new sites will be indexed using Google's
mobile-first indexing - Search Engine Land. [ONLINE] Available
at: https://searchengineland.com/july-1-new-sites-will-be-indexed-using-googles-mobile-
first-indexing-317490. [Accessed 25 September 2019].
Teevan, J., Dumais, S.T. and Horvitz, E., 2005, August. Personalizing search via
automated analysis of interests and activities. In Proceedings of the 28th annual
international ACM SIGIR conference on Research and development in information
retrieval (pp. 449-456). ACM.
Nguyen, T., Rosenberg, M., Song, X., Gao, J., Tiwary, S., Majumder, R. and Deng, L.,
2016. MS MARCO: A Human-Generated MAchine Reading COmprehension Dataset.](https://image.slidesharecdn.com/theuseristhequery-191003145836/85/The-User-is-the-Query-The-Rise-of-Predictive-Proactive-Search-186-320.jpg)
The User is the Query - The Rise of Predictive Proactive Search
- 1. The User is The Query: The Rise of Predictive Proactive Search
- 2. “Today you are you! That is truer than true! There is no one alive who is you-er than you!” (Dr Seuss)
- 4. When introducing Google Feed (now Discover)
- 5. Today’s topic…‘The User is The Query’
- 8. There’s a problem with queries, content & users too
- 9. “In 1998 the web consisted of just 25 million pages…” (Ben Gomez, Google, 2018)
- 10. “… That’s roughly the equivalent number of those in a small library” (Ben Gomez, Google, 2018)
- 11. In 2019… we know the web is huge… billions of web pages (Netcraft, 2019)
- 12. App usage is huge too - By 2018 – App Store has 20 million registered developers. (Techcrunch, 2018)
- 13. 42% of the global population use social media (Emarsys, 2019)
- 14. We are competing with programmatic solutions spraying content & information EVERYWHERE
- 16. Too much choice often has negative impacts
- 17. Almost 98% of visits are people window shopping Average ecommerce conversion +/- 2%
- 18. Despite this… users are still seeking even more information
- 19. Humans forage (like bears) all over the place seeking information… we are informavores
- 20. Researching ALL THE THINGS… before making final decisions
- 21. The number of Google searches increases year on year (Internetlivestat, 2018, curation from various sources)
- 22. 15% of queries every day are new (Google)
- 23. We have become very good at filtering out things which are NOT interesting enough (8 second filter)
- 24. It’s not a short attention- span thing
- 25. Otherwise we would not binge on ‘Stranger Things’
- 26. This is cognitive load management & information filtering
- 27. AT THE SAME TIME words are problematic. Ambiguous… polysemous… synonymous
- 28. In spoken word it is even worse because of homophones and prosody
- 29. Like “four candles” and “fork handles”
- 30. Which does not bode well for the likes of conversational search
- 31. In query understanding sometimes users don’t know what they want either
- 32. Sometimes exactly the same users express an information need in a different way
- 33. Sometimes different users use lots of different ways to mean exactly the same thing
- 34. 'The Vocabulary Problem’ Furnas, G.W., Landauer, T.K., Gomez, L.M. and Dumais, S.T., 1987. The vocabulary problem in human-system communication. Communicatio ns of the ACM, 30(11), pp.964- 971. 1987
- 36. One of the inventors of ‘Latent Semantic Indexing’, created to solve ‘The Vocabulary Problem’ whilst researching at Bellcore (1990)
- 37. BTW… No-one said LSI was used by Google (aside)
- 38. Often words have multiple meanings. Like “like” can be 5 possible parts of speech (POS)
- 39. Sometimes the searcher query is a ‘cold start’ query
- 40. Broad or cold start queries might call for result diversification due to lack of intent detection
- 41. Search engines may return a broad blend of results to match these queries Freshness Serendipity Novelty Diversity
- 43. The searcher has to click around to provide feedback on their intent or reformulate the query by entering something else (‘query refinement’)
- 44. To then deliver sequential queries with greater intent understanding
- 45. Human in the loop
- 46. Query refinement says… “Your move next”
- 47. And maybe… MARKOV CHAINS Where the results are dependent on the previous step https://en.wikipedia.org/wiki/Mark ov_chain
- 48. Like RPG (Role Playing Games) Where choices made determine the next choices given
- 49. People also ask / related queries are a special kind of ‘Query Refinement’
- 50. A kind of ‘probability-driven fork in the road’ http://delivery.acm.org/10.1145/1780000/1772776/p841- sadikov.pdf (Sadikov et al, 2010) CLUSTERING QUERY REFINEMENTS BY USER INTENT
- 51. BUT word’s meaning & user intent /context combined are still very hard to understand for search engines
- 52. Despite huge leaps forward in query classification & natural language understanding
- 53. Accelerated mostly by Google’s BERT
- 56. MS MARCO
- 57. Stanford Question And Answer Dataset 2.0 • Rajpurkar, P., Zhang, J., Lopyrev, K. and Liang, P., 2016. Squad: 100,000+ questions for machine comprehension of text. arXiv preprint arXiv:1606.05250.
- 58. Time (& Space) context shifts query intent further
- 59. The exact same queries have different intent at different times & different locations
- 60. Let’s Take The Query [Easter]
- 61. Modeling & Predicting Behavioural Dynamics on The Web (Radinsky et al, 2012)
- 62. “When users’ information needs change over time, the ranking of results should also change to accommodate these needs.” (Radinsky, 2013)
- 63. This is ‘Query Intent Shift’
- 64. Your ranking flux might well be shifting query intents at scale
- 66. The intent of queries changes over time
- 68. Thought: The Mobile-First Index is built For mobile first and mobile is probably ‘VERY’ query dynamic, local & temporal (shifty)
- 69. The passage of time adds new meaning to queries sometimes too
- 70. The rise and fall of the Blackberry?
- 71. At certain times far more intents will be transactional
- 72. And sometimes only reasons a particular audience would understand spike temporal queries
- 73. [Four candles] + [fork handles] interest over time
- 74. Sometimes it is other events which trigger unexpected queries
- 76. Maybe It’s Time For A Change?
- 77. Enter… The Next 20 Years of Search
- 78. Hmm… That sounds big Google… This is HUGE
- 79. Three FUNDAMENTAL shifts in search
- 80. 1. The shift from answers to journeys 2. The shift from queries to query-less 3. The shift from text to visual information
- 81. The shift from text to more visual information is simple
- 82. This is cognitive load management & information filtering
- 83. Images are much easier to mentally consume than text & audio
- 84. Images & video engage… Images & video entertain Images & video provoke emotion
- 85. Photography app usage had a 210% increase between 2016 and 2018 according to App Annie
- 86. People spend on average 2.6x more time on pages with video
- 87. Image search is curation. Totally different to text-based search
- 88. Think accessibility first with images & videos. You won’t go far wrong (alt / caption / file / description / title)
- 89. Go nuts with quality images & video
- 90. This feels like a huge UX / accessibility shift… Hoorah
- 91. What about the switch to journeys and query-less?
- 92. “Easier if we can model: who is asking, what they have done in the past, where they are, when it is, etc.” (Susan Dumais, CIKM, 2016)
- 93. “Queries Are Difficult To Understand in Isolation” (Susan Dumais, Microsoft Research, 2016)
- 94. AKA - Contextual Search = User + Time + Location + Device + Task
- 95. Better still… what about predicting the user’s informational needs to proactively make suggestions
- 96. “Nevertheless, as the world is becoming more mobile-centric, this old-fashioned query-driven search scenario and clickbased evaluation mechanism can no longer catch up with the rapid evolution of user demand on mobile devices.” (Song and Guo,2016 (Microsoft Research))
- 97. “”Therefore,a more user-friendly, mobile-centric and scenario driven search paradigm that requires minimal user inputs is ready to come out” (Song and Guo,2016 (Microsoft Research))
- 98. Zero-Query Queries – No Query Required
- 99. Personalising Search via Interests & Activities 2005 paper awarded the 2017 SIGIR Test of Time Award. Cited 1029 times to date Teevan, J., Dumais, S.T. and Horvitz, E., 2005, August. Personalizing search via automated analysis of interests and activities. In Proceedings of the 28th annual international ACM
- 100. QueryLess: Next Gen Proactive Search And Recommender Engines (2016)
- 101. It kind of sounds like Google Discover
- 102. At last announcement Google Discover had 800 million users (May, 2018)
- 103. And it’s on the home page of the mobile browser now too... How many users now?
- 104. It knows you… and the things you do… where you’ve been… where you’re going
- 105. Mobile Device Sensors (14 sensors or more) Proximity sensors GPS sensor Ambient light sensor Accelerometer Compass Gyroscope Back illuminated sensor
- 106. ‘The User (needs) is The Query’
- 107. This is Task-Driven Search & Recommender Systems
- 109. Google Scholar is now a Recommender System Too
- 110. QueryLess: Next Gen Proactive Search And Recommender Engines
- 111. “Patterns were spotted about repetitive task driven search behaviours – predictable” (Song & Guo, 2016)
- 112. “Predictable task timeline patterns are more prevalent on mobile devices” (Song & Guo, 2016)
- 113. Like e.g. ‘checking the stock market’ every morning if you’re interested in stocks and shares
- 114. People are creatures of habit it seems
- 115. “In many cases predicting informational needs removes the need for the query & reactive search engine” (Song & Guo, 2016)
- 116. Google Discover looks to be focusing on hobbies, interests, news and social activities
- 118. An information need is rarely a task with a single finite item
- 119. It’s more like a series of little chunks (sub-tasks)
- 120. Tasks & timelines go hand in hand… it seems
- 121. Many tasks & intents can be modelled according to predicted patterns
- 122. Very Recent Microsoft Research
- 123. The Ideal is Personalisation • Not easy to achieve fully • Sparsity of data • Privacy concerns • Broken sequences
- 124. In the absence of personalization… collaborative Filtering
- 125. There are other people nearly like you
- 126. You (and me) are unique… but may be similar
- 127. Matrix Factorisation (Netflix Recommendation System) + Matrix Factorisation (WALS Algorithm, Tensorflow)
- 129. Based on users liking the same things (with hidden common preferences)
- 130. Those sharing similar interests likely share other hidden interests too (i.e. the system does not know of them yet)
- 131. Understand the user, understand their cohort… Understand other similar informational needs
- 133. Modelling cohorts
- 135. Reinforcement learning thrives from rewards (implicit feedback)
- 137. YouTube is a Recommender System
- 138. YouTube Feedback Controls is ‘The Human in The Loop’
- 140. Toward a Personal Knowledge Graph
- 141. The two sides of assistant will both be proactive Provide answers / search Conversation Search Help with activities / tasks Conversation Actions
- 142. Extend Actions on Google using Machine Learning
- 143. Understand your customers to assist with AI Perceived Information need Micro-task Micro-task Micro-task Micro-task Micro-task Task Micro-task Micro-task Micro-task Micro-task Task Micro-task Micro-task Task Micro-task Micro-task Micro-task Task Micro-task Micro-task Task Micro-task Task We can identify the user’s probable top tasks & subtasks Identify their needs & what info they need along the way
- 144. Tell us about the tasks, order and steps involved in booking a hotel
- 145. Many built-in intents & many ‘coming soon’
- 147. Multi-platforming • Switching between search and video • Between search and a recommender system
- 149. Truly PERSONAL AI is not possible without a PERSONAL KNOWLEDGE GRAPH (Krisztian Balog, ECIR 2019)
- 151. A Recent Microsoft Personal Knowledge Graph Patent
- 152. Assistant + Home + Discover + Search App + Desktop + Location Tracker + Calendar + Gmail + YouTube
- 153. In your car
- 154. In your console
- 156. Where the user is truly ‘the query’
- 157. Toward An Audience of One
- 158. What Can SEOs Do About This?
- 159. Realise… your ranking tools are mostly wrong
- 160. Identify interests & affinity groups
- 162. Map every single informational need sub- task you can think of to the sections of a model like the RACE model
- 163. Build task timeline clusters
- 164. Map & cluster ‘Related’ content by task type. Categories are too broad, and topics may be too
- 166. Instead of continually creating Moooaaaarrr content, make what you have better
- 167. Continually update and improve on solid URL evergreen content
- 168. Continually update and improve on solid URL seasonal temporal content
- 169. Map content clearly to tasks & task timelines
- 170. Maximise the ‘Local’ Opportunity
- 172. Think “CRM for SEO”
- 173. Identify predictable patterns of user behavior
- 174. The QueryLess change will not come overnight… things move slowly
- 175. Take-aways
- 176. Go • Go big on evergreen content & keep updated Optimise • Optimise images well – think curation / collections Map • Map user journeys to content plans Optimise • video well – enhance with markup / transcription Get • Get personal – keep refining segments / personas Identify • Identify & cluster content around task timelines Use • Use relatedness across content, tasks & temporality
- 179. References
- 180. You may need a dual or multi- armed content strategy 2%
- 181. Book hotel intent When do you want to stay? dates dates How many nights? 3 nights 2 nights Overnight A week Single or double room? Single room Double room Programme your own expected questions and answers
- 182. References • Broder, A., 2002, September. A taxonomy of web search. In ACM Sigir forum (Vol. 36, No. 2, pp. 3-10). ACM. • Chuklin, A., Severyn, A., Trippas, J., Alfonseca, E., Silen, H. and Spina, D., 2018. Prosody Modifications for Question-Answering in Voice-Only Settings. arXiv preprint arXiv:1806.03957. • HigherVisibility. 2018. How Popular is Voice Search? | HigherVisibility. [ONLINE] Available at: https://www.highervisibility.com/blog/how-popular-is-voice-search/ • Filippova, K., Alfonseca, E., Colmenares, C.A., Kaiser, L. and Vinyals, O., 2015. Sentence compression by deletion with lstms. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (pp. 360-368). • Filippova, K. and Alfonseca, E., 2015. Fast k-best sentence compression. arXiv preprint arXiv:1510.08418. • Google Developers. 2018. Content-based Actions | Actions on Google | Google Developers. [ONLINE] Available at: https://developers.google.com/actions/content- actions/. [Accessed 18 June 2018]
- 183. References • Mitkov, R., 2014. Anaphora resolution. Routledge. • NLP Department - Stanford University - Imran Q Sayed. 2018. Issues in Anaphora Resolution. [ONLINE] Available at: https://nlp.stanford.edu/courses/cs224n/2003/fp/iqsayed/project_report.pdf. [Accessed 28 June 2018]. • Radlinski, F. and Craswell, N., 2017, March. A theoretical framework for conversational search. In Proceedings of the 2017 Conference on Conference Human Information Interaction and Retrieval (pp. 117-126). ACM. • Schalkwyk, J., Beeferman, D., Beaufays, F., Byrne, B., Chelba, C., Cohen, M., Kamvar, M. and Strope, B., 2010. “Your word is my command”: Google search by voice: a case study. In Advances in speech recognition (pp. 61-90). Springer, Boston, MA. • SISTRIX. 2018. Stepping out of the SEO Bubble - SISTRIX. [ONLINE] Available at: https://www.sistrix.com/blog/stepping-out-of-the-seo-bubble/. [Accessed 16 June 2018]. • Presentation at ESSIR2017 on work by Radinsky, K., Svore, K.M., Dumais, S.T., Shokouhi, M., Teevan, J., Bocharov, A. and Horvitz, E., 2013. Behavioral dynamics on the web: Learning, modeling, and prediction. ACM Transactions on Information Systems (TOIS), 31(3), p.16.
- 184. References • The Stanford Question Answering Dataset. 2018. The Stanford Question Answering Dataset. [ONLINE] Available at: https://rajpurkar.github.io/SQuAD-explorer/. • Trippas, J.R., Spina, D., Cavedon, L., Joho, H. and Sanderson, M., 2018. Informing the Design of Spoken Conversational Search. • https://medium.com/@ashishgupta031/sequence-aware- reinforcement-learning-over-knowledge-graphs-a8af155e716c • Jansen, B.J., Booth, D.L. and Spink, A., 2008. Determining the informational, navigational, and transactional intent of Web queries. Information Processing & Management, 44(3), pp.1251- 1266.
- 185. References Radinsky, K., Svore, K.M., Dumais, S.T., Shokouhi, M., Teevan, J., Bocharov, A. and Horvitz, E., 2013. Behavioral dynamics on the web: Learning, modeling, and prediction. ACM Transactions on Information Systems (TOIS), 31(3), p.16 Sadikov, E., Madhavan, J. and Halevy, A., Google LLC, 2013. Clustering query refinements by inferred user intent. U.S. Patent 8,423,538. Official Google Webmaster Central Blog. 2019. Official Google Webmaster Central Blog: Rolling out mobile-first indexing . [ONLINE] Available at: https://webmasters.googleblog.com/2018/03/rolling-out-mobile-first- indexing.html. [Accessed 25 September 2019]. Zhou, S., Cheng, K. and Men, L., 2017, April. The survey of large-scale query classification. In AIP Conference Proceedings (Vol. 1834, No. 1, p. 040045). AIP Publishing.
- 186. References Search Engine Land. 2019. Starting July 1, all new sites will be indexed using Google's mobile-first indexing - Search Engine Land. [ONLINE] Available at: https://searchengineland.com/july-1-new-sites-will-be-indexed-using-googles-mobile- first-indexing-317490. [Accessed 25 September 2019]. Teevan, J., Dumais, S.T. and Horvitz, E., 2005, August. Personalizing search via automated analysis of interests and activities. In Proceedings of the 28th annual international ACM SIGIR conference on Research and development in information retrieval (pp. 449-456). ACM. Nguyen, T., Rosenberg, M., Song, X., Gao, J., Tiwary, S., Majumder, R. and Deng, L., 2016. MS MARCO: A Human-Generated MAchine Reading COmprehension Dataset.
- 187. Appendix
- 188. The task (query mission) tied to the page type matters hugely
- 189. Understand the shared preferences, learn the hidden preferences
- 190. Bias & Reproducibility is a Challenge
- 191. Bias Considerations Presentation Bias Programming Bias Audience Manipulated Bias (e.g fake reviews) Machine Learning / AI Bias (Black box algorithms) Matthew’s Law Zipfian Distribution of Web Content
- 192. Spotify add novelty items to home page to avoid biased personalisation
- 193. Do yourself a favour and follow Mounia Lalmas @mounialalmas
- 194. And this polar bear
- 195. Bias on the web and recommender systems
- 196. NoBIAS Project
- 197. Reproducibility problems in research & RecSys (very high)
- 198. MS MARCO Paper • Nguyen, T., Rosenberg, M., Song, X., Gao, J., Tiwary, S., Majumder, R. and Deng, L., 2016. MS MARCO: A Human- Generated MAchine Reading COmprehension Dataset.
- 199. Query Classifications perhaps? - There are some we know of already
- 200. A Taxonomy of Web Search (Broder, 2002) Informational Navigational Transactional
- 201. Google’s Quality Raters Guide simplifies & extends these Know query == Informational Website query == Navigational Do query == Transactional Visit in person == Local intent
- 202. There are also several types of queries too (Krisztian Balog, ECIR, 2019) Keyword queries (Normal keyword queries) Keyword++ queries (Faceted / filtered queries) Zero-Query queries (User is the query) Natural language queries Structured queries (e.g. SQL)
- 203. 80% of all queries are information al in nature (Jansen et al, 2008) 80% 10% 10% Query Intent Split Informational Transactional Navigational
- 204. Temporal Dynamic Intent (Burstiness) is a huge factor for intent
- 205. Broder et al 39 - 48% Informational 20 – 25% Navigational 30 – 36% Transactional
- 206. “dresses”, “shoes”, “bags” “buy dresses”, “buy shoes”, “buy bags”, “dress sales”, “shoe sales” Really means
- 208. Google now processes over 40,000 search queries every second on average (Source: Dubious Internetlivestats estimation)
- 210. Another Great ‘Ronnies’ Sketch BTW
- 211. ‘iPhone’ – Query Example (Google Quality Raters Guidelines)
- 212. IR researchers competing over natural language understanding
- 213. What did you really mean when you searched for ‘Easter’? • Radinsky, K., Svore, K.M., Dumais, S.T., Shokouhi, M., Teevan, J., Bocharov, A. and Horvitz, E., 2013. Behavioral dynamics on the web: Learning, modeling, and prediction. ACM Transactions on Information Systems (TOIS), 31(3), p.16. When did you search for ‘Easter’? A few weeks before Easter A few days before Easter During Easter What you mostly meant When is Easter? Things to do at Easter What is the meaning of Easter?
- 214. The problem is consistent high precision is nowhere in sight
- 215. And if we are to move into multi-device ubiquitious search then…
- 216. Accuracy of results is more important than quantity
- 217. Paraphrase handling on ‘Actions’ appears to be programmable
- 218. Gets Round The Vocabulary & Disambiguation Problem
- 220. 2017 2017 2017
- 221. Understand your customers to assist with AI Customer Service Data Customer Panels Email questions FAQs Build Assistant App
- 222. Dialogflow
- 223. Extend Actions on Google using Machine Learning
- 224. Think about the different features which matter to users more dependent on the domain News (freshness) Jobs (salary, job title, location) Restaurants (location, cuisine) Shopping (price)
- 225. By 2022 PCs will account for only 19 percent of IP traffic (Comscore, 2019)
- 226. We know there are various versions of query intent classifications (Broder, 2002; Rose; Jansen, 2008)
- 228. Interest over time for Google Home & Amazon Alexa
- 229. App usage is also huge