Wednesday, July 31, 2013
Tuesday, July 30, 2013
Monday, July 29, 2013
Monday, July 22, 2013
Bing vs Google suggestions
Spot the difference on {royal baby} - which on has better suggestions the unbranded search engine or the other one?
Friday, July 19, 2013
What they've said about new Knowledge Suggestions @ Bing
Other articles:
Bing Autosuggest Categories Revamped
UberGizmo (blog)
Microsoft expands Bing’s autosuggest for people to include animals, brands, movies, albums, places, and more
thenextweb.com
Marketing
Pilgrim
MediaPost
IN
Wednesday, July 17, 2013
Disrupting the searchbox
Today,
I published a blog on Microsoft search and it has been quoted by many
I published a blog on Microsoft search and it has been quoted by many
“This
is one area where Bing has a clearly better user experience than Google, if
you’re looking to get actionable results from the actual search box. While
Google will put a Pitbull Google+ profile in its autosuggest feature, it does
little with suggestions for any of the examples Bing provides.
That’s
not to say Google’s Knowledge Graph results don’t provide ways for users to
distinguish what they’re looking for, but you have to get through the search
query before you can tell Google what you’re actually looking for (like the
Harry Potter books vs. movies, for example). Of course, you could always
specify in your actual query.”
Sunday, July 14, 2013
SQL-like engines on the top of hadoop map/reduce
Thursday, July 11, 2013
Building large scale systems at Twitter
Very interesting article for building large systems at Twitter. Worth reading it, use of sharding on the top of standard mysql makes the trick (gizzard - for sharding , flockdb - for graph db)
https://twitter.com/zipkinproject
http://www.youtube.com/watch?v=5cKTP36HVgI
http://highscalability.com/blog/2011/12/19/how-twitter-stores-250-million-tweets-a-day-using-mysql.html
http://highscalability.com/blog/2013/7/8/the-architecture-twitter-uses-to-deal-with-150m-active-users.html
https://twitter.com/zipkinproject
http://www.youtube.com/watch?v=5cKTP36HVgI
http://highscalability.com/blog/2011/12/19/how-twitter-stores-250-million-tweets-a-day-using-mysql.html
http://highscalability.com/blog/2013/7/8/the-architecture-twitter-uses-to-deal-with-150m-active-users.html
Tuesday, July 9, 2013
Monday, July 8, 2013
Othello
Othello is a game played on a square board, with black and white pieces. If a white/black piece is surrounded by a black/white one (either left/right or up/down) then the piece will change color. Players play in turn until they have valid moves left. Implement Othello following Object Oriented criteria (e.g. what is a piece, what is a board, what is a game, who keeps the score, what is a valid move).
represent a double in [0,1) as binary
First need to understand what is the binary representation for a double in [0,1). Let's see an analogy. If the number is an integer such as 123 I would probably make something like this
2 ^ 7 = 128 > 123 too large
123 - 2 ^ 6 = 59 -> set bit 1
59 - 2 ^ 5 = 27 -> set bit 1
27 - 2 ^ 4 = 11 -> set bit 1
11 - 2 ^ 3 = 3 -> set bit 1
3 - 2 ^ 2 < 0 -> set bit 0
3 - 2 ^ 1 = 1 -> set bit 1
1 1 1 1 0 1 1
SearchCharm - Unified Search Experience in Windows 8.1
I am using windows 8.1 preview since many weeks without any problem. Search experience brings together the online and the offline worlds and it allows users to look for information virtually everywhere.
Traditional Web Search
Do I want to access cnn? Easy. I start to insert the query and I get both the web site and some relevant suggestions including cnn news and cnn money. Suggestions are generated by my group in London.
Navigating windows
Do I want to access windows control panel? No problem, I just use the same interface and search paradigm.
Cloud search
Local search
Traditional Web Search
Do I want to access cnn? Easy. I start to insert the query and I get both the web site and some relevant suggestions including cnn news and cnn money. Suggestions are generated by my group in London.
Do I want to access windows control panel? No problem, I just use the same interface and search paradigm.
Cloud search
What if I want to access my files stored in the cloud? Here I have my cv saved on Microsoft Skydrive and accessing it is as easy as searching via the charm
What if I want to access my files stored in my local computer? Here I have some videos saved on my local computer and accessing them is as easy as searching via the charm
Sunday, July 7, 2013
Given a very large binary tree T1 and a small binary T2
check whether T2 is contained in T1. Discuss your solution
Saturday, July 6, 2013
Given a binary tree, create a list of lists containing all the nodes in one level
write some c++ code / STL. Optimal in space and time
Friday, July 5, 2013
Wednesday, July 3, 2013
Search as you type with snapshots and disambiguation
Exciting day today! Bing is shipping an innovative search feature aiming at helping users to complete their tasks, as soon as they start to type their queries.
Another example is the query {pluto}. Is the user interested in the planet or is she interested in the cartoon? Those are very different kinds of entities which happens to have the same name but Bing understands the difference and gives to the user the opportunity to select the desired choice.
Now let's suppose that the intent is not ambiguous. For instance consider the keyword {michele obama}, we believe that it is very valuable to have a rich suggestion like the following one.
And what if there are two persons with the same name? In this case, we offer the opportunity to select the desired person as in this example for the query prefix {rachel nich}
One more example. Let's consider the query {u2}. Here, we offer the opportunity to explore all the entities / artists that are playing with the pop band
Hope you enjoy it - this is the beginning.
Consider the keyword "pitbull". This is a very interesting query because the user can be interested in getting either results about the famous singer (a person) or about the dog (an animal). For this example, we believe no instantaneous decision can be taken on the user's behalf and that only the user has a complete understanding of the intent expressed with a query.
However, we believe that the search engine can help the user to express very quickly this intent. How? In Bing we use the power of Satori (our Microsoft Knowledge Graph) to enrich our suggestions with additional information useful for disambiguating different meanings associated with the query. In this way, the users can select their own most appropriate choice.
However, we believe that the search engine can help the user to express very quickly this intent. How? In Bing we use the power of Satori (our Microsoft Knowledge Graph) to enrich our suggestions with additional information useful for disambiguating different meanings associated with the query. In this way, the users can select their own most appropriate choice.
Another example is the query {pluto}. Is the user interested in the planet or is she interested in the cartoon? Those are very different kinds of entities which happens to have the same name but Bing understands the difference and gives to the user the opportunity to select the desired choice.
Now let's suppose that the intent is not ambiguous. For instance consider the keyword {michele obama}, we believe that it is very valuable to have a rich suggestion like the following one.
And what if there are two persons with the same name? In this case, we offer the opportunity to select the desired person as in this example for the query prefix {rachel nich}
One more example. Let's consider the query {u2}. Here, we offer the opportunity to explore all the entities / artists that are playing with the pop band
And what if you search for {harry potter}? In this case Bing provides a snapshot of the filmography, a reference to the book and a reference to the character.
Hope you enjoy it - this is the beginning.
Subscribe to:
Posts (Atom)