Upgrade to Pro — share decks privately, control downloads, hide ads and more …

HR-Tech Meetup: Voice Controlled Job Search

Avatar for Potentialpark Potentialpark
February 07, 2018

HR-Tech Meetup: Voice Controlled Job Search

A discussion topic for the HR-Tech Meetup in Stockholm, Feb 7 2018.

Avatar for Potentialpark

Potentialpark

February 07, 2018
Tweet

More Decks by Potentialpark

Other Decks in Research

Transcript

  1. © Potentialpark 2018 State of the art “Siri, find me

    jobs for full-stack developers in Stockholm”
  2. © Potentialpark 2018 State of the art •  “Google voice

    search queries in 2016 are up 35x over 2008” - Google trends via Search Engine Watch •  “In May 2016, 1 in 5 searches on an Android app in the USA were through speech” - KPCB •  “56% of online grocery shoppers use or plan to use voice controlled smart assistant/speaker” - Global Web Index
  3. © Potentialpark 2018 The candidates’ voice Would you find it

    helpful to use voice control for career searching purposes?
  4. © Potentialpark 2018 Intro Is it possible today to search

    for jobs by using our voice? How? Yes! AI can listen to us and get back with a list of matching jobs
  5. © Potentialpark 2018 Intro What do we need? •  Speech

    Recognition API •  Artificial Intelligence and Machine Learning •  Job Search API
  6. © Potentialpark 2018 Keywords •  API (open) –  App which

    allows developers to access backend data and use them for their own applications. •  Speech recognition –  Machines translate a spoken language (voice) into text –  Cloud Speech, Bing Speech, Watson Speech and others
  7. © Potentialpark 2018 Keywords •  Artificial Intelligence –  Device that

    takes actions by maximising the chance of success –  More simply, a machine that solves problems in a ‘smart’ way •  Machine Learning –  Ability of a machine to learn without being programmed.
  8. © Potentialpark 2018 Scenario Purpose: •  To find jobs in

    Stockholm for Developers via voice •  Get data from a known source of jobs
  9. © Potentialpark 2018 Scenario Solution: •  Speech recognition service that

    translates voice into text •  Build an AI who can handle human to machine conversations (i.e. Dialogflow, owned by Google) •  Webhooks for the fulfillment
  10. © Potentialpark 2018 How to talk to AI? •  Invocation

    •  Intent •  Fulfillment •  Response
  11. © Potentialpark 2018 1. Invocation Invoke the App by a

    command predefined by the developer ‘Hello, Helena!’
  12. © Potentialpark 2018 2. Intent 1/5 A set of permutations

    of the same question helps the AI to better understand the request of the user
  13. © Potentialpark 2018 2. Intent 2/5 Role and location are

    mandatory entities for our specific case
  14. © Potentialpark 2018 2. Intent 4/5 Let’s say to the

    agent the following sentence: “Please show me jobs as software engineer in Stockholm” What will it be the agent’s answer?
  15. © Potentialpark 2018 3. Response •  Use web hooks to

    send ‘role’ and ‘location’ to a Job Search API •  Provide the user with the list of jobs: –  Display a list and let the user read it or (probably) better.. –  Use a Text-to-Speech API and let the user listen!
  16. © Potentialpark 2018 Conclusions Who are the biggest providers of

    a Job Search API? Will your candidates be able to see your job ads by using Helena?
  17. © Potentialpark 2018 Thank you! Torgil Lenning Founder and chairman

    of Potentialpark #potentialparkpeople Stefano Chialà Software Engineer at Potentialpark