How International Image/Video Search Works (Heba Ashour and Jason Barnard)

Heba Ashour talks with Jason Barnard about how international image/video search works.

Heba Ashour is Principal Program Manager Lead of Multimedia Search Ranking and Platform at Microsoft Bing. She is an experienced leader, technology manager, technical visionary, and passionate about using technology to democratise access to information, services, and opportunities. Heba focuses on providing users of search services around the world with engaging images/videos that match their exact intent in every scenario, empowering them to be informed, entertained, save money, and succeed. Previously, she worked at Microsoft Research’s Advanced Technology Lab in Cairo, where she turned various research technologies into features that were integrated into various Microsoft products.

In this delightful episode, Heba Ashour and Jason Barnard discuss how search engine users around the world are increasingly using visual and video search to get a quick and accurate answers to their questions and solutions to their problems. Heba explains that Bing’s massive technology advances and overcoming bandwidth and technology limitations have vastly increased their capacity to deliver the most relevant results for users’ search queries, regardless of the language they use. And that drives user demand, expanding expectations, and an increasingly immersive, multimedia user experience on search engines like Bing and Google.

That said, even in the modern world of Search (circa 2022), text search is still heavily used… but to what extent is image and video search really making a difference, is it replacing or supplementing text searches and how do the two interact? Then, as people get busier and busier in this fast-paced world, and as search engines like Bing get better exponentially at understanding the contents, will we see an increasing shift towards image and video search.

As a wonderful bonus, towards the end you’ll learn more about the importance of human intelligence and human annotators in machine learning… and do the algorithms have aesthetic appreciation of the images they are showing, and to what extent does that play into the results we see?.

As always, the show ends with passing the baton… Heba sweetly hands over to next week’s incredible guest, Mike Ulmer.

What you’ll learn from Heba Ashour

00:00 Heba Ashour and Jason Barnard
00:39 Heba Ashour’s Brand SERP from Bing
03:20 Are We Shifting Away From Text Search to Image/Video Search?
04:40 Bing Overcoming Bandwidth and Technology Limitations
06:23 How International Language Affects Search
07:30 What Does Bing Use for Indexing?
09:11 How is Multi-Modal Represented in Semantic Space?
10:27 What Does Semantic Space Mean?
11:29 Will the Vector Space for a Given Semantic Term be Similar in Different Languages?
12:17 How Does Culture Influence Textual Representation?
13:33 Language Agnostic and Language Specific Aspects in Image/Video
14:22 The Importance of Human Intelligence and Human Annotators in Machine Learning
17:03 Human Annotations VS Automated Annotations
24:57 Do Figure and FigCaption Tags (Semantic HTML5) Help Images and Video SEO?
30:00 Do Publishers’ Transcripts and Subtitling Help SEO?
34:45 How Do Search Engines Understand Aesthetically Pleasing Images?
36:30 The Concept of Super Resolution for Images
38:39 Passing the Baton: Heba Ashour to Mike Ulmer