Next Question
I wonder how many times he's heard the joke, "you've wasted your time, the answer is always 42"?
In my mind what Wolfram has accomplished is the natural progression of search engines.
Since September 1998 (when Google Inc opened its doors) {1}, search engines have worked to catalog (spider) and categorize a large number of web pages. The goal has been to answer user queries with the most relevant results from cataloging and categorization processes.
As the Internet has grown and this network has defined itself as "the message" {2} search engines have become increasingly important. Internet industry leaders (such as yourself) have accurately predicted that the public needs access to relevant, correct and importantly well supported information.
If Wolfram's system works as it is demonstrated and described, the major breakthroughs lie in both the natural language input and of course the answers that the system is capable of providing.
While it's still very early, I think that the biggest hurdle the system will need to overcome is the quality of the information itself. The content that the system is fed (or finds) still needs to be accurate or at the very least verifiable against other sources. This drawback will not be immediately prevalent because as Mahalo itself has found, the answers to popular questions are much easier to reference then questions that rely either on very specific information or niche subjects.
The world will be watching Wolfram. This system has the capability to redefine not only searching online, but the content and in turn the network itself.
A few other sources of information:
http://news.cnet.com/wolfram-alpha-next-major-search-breakthrough/
http://blogs.computerworld.com/stephen_wolfram_alpha_snake_oil_or_skynet
http://www.semanticuniverse.com/blogs-i-was-positively-impressed-wolfram-alpha.html
Source(s):
{1} http://www.google.com/corporate/history.html
{2} http://twitter.com/mcuban says "The lessons of twitter, facebook and social networks..The Medium is No Longer the Message. The Network defines the message"
Permalink | Report
I watched about 15 minutes of the video, and it's hard to follow because all you mostly see is him talking and typing something, but you don't see what exactly he types, and you don't see what the app is coming back with. Also he's not saying anything much about how it works.
From the piece that I watched, it seems like it'll be a very useful where there are very well organised and trustworthy datasets that it has been told to use, and someone has figured out some typical types of queries people would have, and how to "compute" the answers to those.
It's a little bit reminiscent of AI systems that can look very good in demos but it often turns out that it takes a huge amount of work to get the thing to work that well for that demo, and it doesn't scale up well to put it mildly.
I'd love it if this was a big breakthrough, but I haven't seen anything so far that convinces me that it really is.
Permalink | Report
But from what I can understand, it seems like a very interesting project.
Part of my research in the past was on building tools for domain experts to describe their data in a computer interpretable form and to use this method in many ways as a replacement for trying to get AI/Text Mining tools to figure it out automatically.
This idea always met with great skepticism. The general complaint being that there is too much data out there and it's too much work for humans to get involved.
From what it sounds like Mr. Wolfram is setting up WolframApha to use mostly human curated data.
This is what interests me most, along with the idea of matching the available answers to the user's question and not the other way around.
I will be most interested to see if this system becomes open for outside research and development or becomes another secretive monolith like Google. I'm hoping for at least some openness.
And probably most important is what breakthroughs they have made in the process of curating data.
So final answer, very interesting stuff!!
Thanks for introducing it to us.
Wikipedia page for those without the time to watch all the video:
http://en.wikipedia.org/wiki/Wolfram_Alpha
Permalink | Report
The internet is a wild and uncontrolled place. I think Mahalo has the right idea, hand tuned search results that provide added value. Let the machine wade through the mountain of data and the human can polish the ore they unearth.
One hour and forty five minutes for a demo is to long. If you can't show me in 5 minutes how great your product is you need to do more work. When I was first introduced to Google it was obvious that it was better than anything available within a few minutes of using it.
Permalink | Report
Answered Question
Best Answer Decided by Votes
| April 30, 2009 01:34 AM |
In my mind what Wolfram has accomplished is the natural progression of search engines.
Since September 1998 (when Google Inc opened its doors) {1}, search engines have worked to catalog (spider) and categorize a large number of web pages. The goal has been to answer user queries with the most relevant results from cataloging and categorization processes.
As the Internet has grown and this network has defined itself as "the message" {2} search engines have become increasingly important. Internet industry leaders (such as yourself) have accurately predicted that the public needs access to relevant, correct and importantly well supported information.
If Wolfram's system works as it is demonstrated and described, the major breakthroughs lie in both the natural language input and of course the answers that the system is capable of providing.
While it's still very early, I think that the biggest hurdle the system will need to overcome is the quality of the information itself. The content that the system is fed (or finds) still needs to be accurate or at the very least verifiable against other sources. This drawback will not be immediately prevalent because as Mahalo itself has found, the answers to popular questions are much easier to reference then questions that rely either on very specific information or niche subjects.
The world will be watching Wolfram. This system has the capability to redefine not only searching online, but the content and in turn the network itself.
A few other sources of information:
http://news.cnet.com/wolfram-alpha-next-major-search-breakthrough/
http://blogs.computerworld.com/stephen_wolfram_alpha_snake_oil_or_skynet
http://www.semanticuniverse.com/blogs-i-was-positively-impressed-wolfram-alpha.html
Source(s):
{1} http://www.google.com/corporate/history.html
{2} http://twitter.com/mcuban says "The lessons of twitter, facebook and social networks..The Medium is No Longer the Message. The Network defines the message"
Permalink | Report
Voted as best: maurice, tracebooks
Reply 
Other Answers (3)
April 30, 2009 01:36 AM
I'll be very interested to see what Wolfram Alpha is like when it's opened up for public use. I watched about 15 minutes of the video, and it's hard to follow because all you mostly see is him talking and typing something, but you don't see what exactly he types, and you don't see what the app is coming back with. Also he's not saying anything much about how it works.
From the piece that I watched, it seems like it'll be a very useful where there are very well organised and trustworthy datasets that it has been told to use, and someone has figured out some typical types of queries people would have, and how to "compute" the answers to those.
It's a little bit reminiscent of AI systems that can look very good in demos but it often turns out that it takes a huge amount of work to get the thing to work that well for that demo, and it doesn't scale up well to put it mildly.
I'd love it if this was a big breakthrough, but I haven't seen anything so far that convinces me that it really is.
Permalink | Report
Reply
April 30, 2009 01:40 AM
Very strange they didn't actually cut to the screen shots he's talking about... very confusing to follow, I agree.
Report
April 30, 2009 02:40 AM
Further to my answer... he does talk a lot more about how it works in the Q&A part of the event. And it does entail a lot of work to find reliable data souces, pull in feeds and clean them up, and get domain experts to figure out what kind of queries to do, and how to interpret the language used in queries.
Overall it's pretty darn impressive that anyone's put in that amount of work.
Report
Overall it's pretty darn impressive that anyone's put in that amount of work.
April 30, 2009 02:59 AM
I am about 45 minutes in to the video, and yes the camera work does leave something to be desired. But from what I can understand, it seems like a very interesting project.
Part of my research in the past was on building tools for domain experts to describe their data in a computer interpretable form and to use this method in many ways as a replacement for trying to get AI/Text Mining tools to figure it out automatically.
This idea always met with great skepticism. The general complaint being that there is too much data out there and it's too much work for humans to get involved.
From what it sounds like Mr. Wolfram is setting up WolframApha to use mostly human curated data.
This is what interests me most, along with the idea of matching the available answers to the user's question and not the other way around.
I will be most interested to see if this system becomes open for outside research and development or becomes another secretive monolith like Google. I'm hoping for at least some openness.
And probably most important is what breakthroughs they have made in the process of curating data.
So final answer, very interesting stuff!!
Thanks for introducing it to us.
Wikipedia page for those without the time to watch all the video:
http://en.wikipedia.org/wiki/Wolfram_Alpha
Permalink | Report
Voted as best: jeffhoard
Reply
May 04, 2009 11:01 AM
If nothing else, they are doing a good job of stirring up the press with "the next google talk".
Linked from Huffington post as "Newer, Smarter Search Engine Could Challenge Google's Dominance"
http://www.independent.co.uk/life-style/gadgets-and-tech/news/an-invention-that-could-change-the-internet-for-ever-1678109.html
Report
Linked from Huffington post as "Newer, Smarter Search Engine Could Challenge Google's Dominance"
http://www.independent.co.uk/life-style/gadgets-and-tech/news/an-invention-that-could-change-the-internet-for-ever-1678109.html
April 30, 2009 03:53 AM
Demos are always very tightly controlled. If you can't make your product look good in a demo then you are really in trouble. I wish Wolfram Alpha the best but I'll reserve judgment until I see the product in use. Relevant search is a hard problem, Cuil.com promised much and failed to deliver. The internet is a wild and uncontrolled place. I think Mahalo has the right idea, hand tuned search results that provide added value. Let the machine wade through the mountain of data and the human can polish the ore they unearth.
One hour and forty five minutes for a demo is to long. If you can't show me in 5 minutes how great your product is you need to do more work. When I was first introduced to Google it was obvious that it was better than anything available within a few minutes of using it.
Permalink | Report
Reply
April 30, 2009 04:27 AM
You make some good points.
Although I don't really see Mahalo as too much of a direct competitor or better alternative.
I see them as dealing with very different types of information.
In many ways Mahalo and WolframAlpha could compliment each other.
But as you said, it remains to be seen if it will actually work.
And some better sources of information are needed.
Also, I'm interested to hear Jason's opinion on WolframAlpha.
Report
Although I don't really see Mahalo as too much of a direct competitor or better alternative.
I see them as dealing with very different types of information.
In many ways Mahalo and WolframAlpha could compliment each other.
But as you said, it remains to be seen if it will actually work.
And some better sources of information are needed.
Also, I'm interested to hear Jason's opinion on WolframAlpha.
April 30, 2009 03:37 PM
It's not a demo in that sense. It's a talk at Harvard to a bunch of people that are interested in the topic. Like an open seminar.
The demo part is about 20-30 mins, the rest is Q&A.
Report
The demo part is about 20-30 mins, the rest is Q&A.
Answer this Question
Related Questions
after performing poorly in my first semester of college, i've started to think about ...
Are you a better talker or a better listener...which is the more valuable 'talent'? I...
Will sending lower ACT scores along with higher SAT scores hurt me in the admissions ...
Would you pose naked on a poster if you thought it would save a few animals from bein...
Are you a better talker or a better listener...which is the more valuable 'talent'? I...
Will sending lower ACT scores along with higher SAT scores hurt me in the admissions ...
Would you pose naked on a poster if you thought it would save a few animals from bein...
Ask a Question
Buy Mahalo Dollars with Credit Card or PayPal
Top Members
Most Popular Tags
Categories
- Anonymous
-
Arts & Design
-
Beauty & Style
-
Books & Authors
-
Business
-
Cars & Transportation
-
Consumer Electronics
- Coupons Deals
-
Education
-
Entertainment
-
Environment
-
Fitness
-
Food & Drink
- From Email
- From Iphone
- From Twitter
-
Health
-
History
-
Hobbies
-
Home & Garden
-
How Tos
-
Humor
-
Jobs
-
Legal
-
Local
-
Love & Relationships
-
Mahalo Answers Community
-
Money
-
Music
-
News
-
NSFW
-
Parenting
-
Pets
-
Science & Mathematics
-
Services
-
Shopping
-
Social Science
-
Society & Culture
-
Sports
-
Technology & Internet
-
Travel
-
Video Games
Welcome New Members
- ginie1014, November 11, 2009 05:41 PM
- donabpoppin, November 11, 2009 05:40 PM
- cool5555, November 11, 2009 05:39 PM
- achayan1978, November 11, 2009 05:32 PM
- achayan, November 11, 2009 05:27 PM
Mahalo Dollars are the currency of Mahalo Answers.
Each Mahalo Dollar costs $1.
Once you earn more than 40 Mahalo Dollars, you can request to be paid via PayPal. Each Mahalo Dollar is currently worth $0.75 when paid out via PayPal. Learn More
It's more like a mashup of a high-quality database, natural language query system and Excel. It does not seek answers in the wilds of the web, but from its own tightly controlled data sources. It doesn't try to answer all kinds of questions, only ones that are "computable". It wouldn't try to answer a question like "What are the strengths of Mahalo?" It would (if one of its data sources had the relevant info to compute from) answer a question like "What are the trends for visitors to Mahalo.com?"
It's other major point is that it is not relying on the precise info you want being on a web page. It can take info from it's sources and crunch it to produce it in the form you asked for. ("What are the trends for the number of visitors to Mahalo.com compared to overall net traffic trends?")
So it's more like an automation of a subset of things you might do manually when researching quantitative info about a topic. (The evaluating of potential data sources, fetching the actual data, reworking it into an intelligible form to examine what you're interested in.)
It's designed more for people that want to know things like: "What are the survival rates using drug X versus drug Y?" than "What are some iPhone games I'd like?"
I think that we took many of the same points away from the demo.
Today, Search Engines can accurately catalog and categorize a page of information. The natural progression of Search is no longer in these areas but rather in the the refinement of those processes and importantly how the data is accessed.
It doesn't look like Wolfram has created "the next Google". It looks like he has figured out the best way to ask Google questions.
I am of course over generalizing by saying "Google". The dataset that Wolfram's system uses is as important as the language in which the questions are asked.
"So (Wolfram's system) is more like an automation of a subset of things you might do manually when researching quantitative info about a topic."
It's not much fun to ask questions that you already know the answer to. To be powerful, a natural language search must be able to find the common alongside the unique. The trick really seems to be validating results to ensure that the best and correct information is being output.
http://news.bbc.co.uk/1/hi/technology/8026331.stm