Multimedia Search | Another Great Debate!

July 31st, 2007 by Charles S. Knight
Posted in Debates | No Comments »



This evening we are very fortunate to have three highly respected participants in our Debate on Multimedia search. They are:





PodNova: “your place to find podcasts, videoblogs and your favorite blogs.”


Blinkx: “the world’s largest and most advanced video search engine.”

EveryZing: “the most powerful digital media merchandising platform available today.”


1) Until Google announced Google Universal, search was very vertically oriented. Do you think this trend to integrate multimedia search into one search bar will continue?

Yes. Google’s Universal Search release was less about innovation and more an acknowledgment that users don’t really use vertical search or “tabs.” Having run Terra Lycos’ worldwide search properties, I can tell you that single-digit percentages of users use the vertical tabs. The question of vertical search is really something created out of the industry rather than out of consumer demand. When you think about it from a user’s perspective, vertical search doesn’t make a lot of sense. Videos are on the Web, so why can’t I search them from the “Web Search” box? Users are interested in all content on a given topic, regardless of form factor. They expect the search box to do the magic of finding the right content and organizing it into an intuitive presentation. This puts a large burden on the content producer, regardless of form factor, to optimize all of their content for the search engines.

No, we don’t think that multimedia search will be integrated in one search bar. We approach ’searching’ as an integrated part of the whole consumption process of our users. First you want to find things, then you want to subscribe to or share it. The you download it to play it on a mobile device, or you play it online with onsite players.

– yes and no. While it’s useful to integrate text and multimedia results
when you do a search and are not sure of exactly what you want back, in most cases, users are increasingly focusing their efforts – they know from the outset that they are looking for video, or images, or text – or even encyclopedic-type information. If they are looking for a video of something, why bother with the images, text, etc.

2) How can publishers of multimedia content use search to make their companies more successful and what can they do to differentiate themselves?

The critical step is to be sure that the multimedia content can be found by the big text Web crawlers. This means that if your video doesn’t have a lot of text associated with it, it will be at a disadvantage. The vast majority of search traffic originates from the “Web search” box on the big four search engines (who control more than 90% of all searches). In addition, the ability to scalably create meta information for multimedia content is absolutely critical.

Content is still king. If you have the content that users want, you will end up higher in search results.

This is of paramount importance for them – if users can’t find their
content they won’t watch it. Having a handful of easy-to-find without search clips won’t work because that doesn’t address the long tail requirements of the Internet marketplace. blinkx published a whitepaper on video SEO – with recommendations on how to optimize your multimedia content for search – you can find it here.

3) With the increase in online video search and advertising start-ups, there has been a lot of emphasis on enhancing the user experience. What are you offering to users and your customers to do that?

EveryZing’s biggest innovation is the ability for users to search within videos and jump to the section of a clip based on their search terms. These “snippets” we create can be accessed from both the search results pages, as well as the player landing pages we deliver. Users have grown accustomed to this level of control in their Web search experience, and they desire it on a video search experience as well.

We’re focusing on a full integrated experience for our users. We want to be the place for users to find, subscribe to, share, play and download their podcasts and vodcasts. As we also create a desktop application (available for free for windows, mac and linux) we also provide ways for our users to sync their content with their offline devices like mp3 players etc.

blinkx offers the largest index of video content on the Web today; from an extremely diverse base of content providers – ranging from traditional outlets like Reuters and CBS, to user-generated content like YouTube and Revver, to niche channels like CyclingTV and Ministry of Sound. We’ve made an enormous variety and amount of video content available through a single destination. Also, we’ve built tools like blinkx Remote – which help users navigate directly to the increasing amount of long-form content on the Web – like popular TV shows such as Lost.

4) Online advertising is expected to skyrocket within the next year. What challenges lie ahead for targeted multimedia advertising?

Without being able to understand the content of a video, video ad targeting will remain as a blunt instrument. Tapping the full contents of a video brings multimedia into the world of contextual advertising, which is a business with $15BB in revenue this year. In addition, advertisers need to create more short form Web-purposed video creative to effectively target online video consumers without alienating them with lengthy pre-rolls or repurposed broadcast advertising content.

Correct profiling so targeting will be spot on. If you get this really good, then “advertising” will become “interesting information” for your users.

I think we’ve yet to arrive at a format that truly works – both for
advertisers AND for the user as well. Ideally, online video advertising
should combine the emotive power of TV promotion, with the utility of
contextual search advertising – the experience should be relevant and unobtrusive for users.

The floor is now open for your questions and comments!

test

People Search | Did we miss one? ;-)

July 31st, 2007 by Charles S. Knight
Posted in Alts, Updates | No Comments »

 Wink  ;-) is one of the new people search engines.  They build their own index of people on the Web; so its not a meta search engine.  They also focus on treating profiles differently than just Web pages – paying attention to structures such as name, location, age, school, work, etc.  Wink claims to search over 200,000,000 profiles across MySpace, Bebo, Friendster, Y!360, Live Spaces, Xanga, Wikipedia, IMDB, Twitter, and more.

Wink’s vision is to provide the fastest and easiest way for people to find, learn about and connect with other people across the Web. People searches are increasingly the way people connect.  And Googling people can yield some pretty ‘random’ results, according to Wink ;-) .

At Wink, you can search for old friends, new friends, business contacts, celebrities, blind dates ;-) , and anyone else by name, location, interests, school, work, groups and more. Wink also lets people manage their on-line presence by claiming their own places.

So, did we miss a good one?  Check it out, eh? Nudge, nudge,   ;-)    ;-) 
 

Exploring the Semantic Web with Dr. Watson

July 31st, 2007 by Charles S. Knight
Posted in Alts, Newcomers | No Comments »

It was another oppressively muggy day in London, and Sherlock Holmes and Dr.Watson were lazily discussing a shocking new discovery. “Great Scott, Holmes, ” Watson exclaimed, “they’ve gone and named an Alternative Search Engine after me! But what in the blazes is an RDF? And what do they mean by the Semantic Web?

“Elementary, my dear Watson,” Holmes replied as he reached for his laptop. He navigated to Answers.com. “Here’s your answer.”

RDF: A recommendation from the W3C for creating meta-data structures that define data on the Web. It is designed to provide a method for classification of data on Web sites in order to improve searching and navigation.

RDF is based on XML

RDF is implemented in XML (RDF/XML) and adds new rules to XML as well as reducing some constraints. For example, it supports flat structures, eliminating the need to have data nested in a hierarchy. RDF is structured as one or more Triples: (1) the subject (what the data is about), (2) the property (an attribute of the subject) and (3) the actual value.

RDF is a common denominator

RDF serves as a common denominator that allows different sets of vocabulary to be recognized. For example, using the RDF descriptor for zip code would let systems exchange zip code data that use “ZIP” and “ZIPCODE. See XDI and RSS.

“I’m afraid that’s just a bit over my head, dear fellow. After all, I’m a doctor, not an engineer!” “Here,” Holmes handed Watson a large piece of scrap paper that he had been doodling on, “perhaps this chart will help you.”

Watson’s face was flushed. “But what the devil has all this got to do with me? Where do I come in?” Clearly the good doctor was becoming exasperated.

“Because, my good man, it says right here,” Watson is a gateway for the Semantic Web, which has been guided by the requirements of Semantic Web applications and by lessons learnt from previous systems. Watson exploits the strengths of semantic technologies to provide fundamental functionalities for a more suitable access to online knowledge.

“There now,” Holmes asked, turning away from the screen, “do you understand now, Watson?” Watson’s chair was empty; Holmes was alone in the room! Downstairs he heard the faint thud of the front door slamming shut. “Watson? Watson?”

What is Not a Search Engine?

July 31st, 2007 by Charles S. Knight
Posted in Guest Authors | 5 Comments »

Today we continue our 3 part series on “What is a Search Engine?” with Part II “What is Not a Search Engine?” by Kaila Colbin.

Part I: What is a Search Engine? by Nitin Karandikar (Yesterday)

Part II: What is Not a Search Engine? by Kaila Colbin (Today)

Part III: What is an Alternative Search Engine?* by me (Tomorrow)

*to be followed by the “Top 100 Alternative Search Engines” August Update.

What is not a search engine?

When we began the conversation that eventually led to this piece, our focus was on how to define search engines. What we found, though, is that you can’t discuss the foreground without discussing the background, and you can’t discuss what a search engine is without discussing what it isn’t. That’s where I come in.

Foreground vs background: is this a picture of a man playing sax or woman’s face?

The question of what is and isn’t a search engine came up for more than one reader on Charles’ ‘Day without Google’, when people were invited to spend 24 hours without accessing the top 5 search engines. Nick T had this question:

Which ones count as “Meta Search” engines? I know many are listed in the top 100 as MetaSearch (as opposed to image/blog/etc), but it seems like just about all of them have at least some results from places like Ask and MSN at the very least. For example, would Clusty count as “Meta”?

A ‘meta search’ engine is one that searches search engines. So because they aren’t generating the results themselves but rather compiling results generated by others, are they search engines or not?

Other participants in the Day without Google used Quintura, which offers a different interface for results from the majors. So is Quintura a search engine or isn’t it?
Read the rest of this entry »