Slashdot Mirror
Is Tableau The Next Google?
Roland Piquepaille writes "At least, the founders of Tableau Software, a small company established in 2003 and based in Seattle, come from Stanford University, where they worked down the hall with Google founders Larry Page and Sergey Brin back in 1997. In 'Tableau making name for itself,' the Seattle Post-Intelligencer writes that Tableau intends to make structured databases easy to use the way Google did with unstructured data. So the company is turning databases into easy-to-generate graphics. Tableau doesn't say who are its customers, but claims that it has more than 100 installations and that it's already profitable. This graphical data mining tool runs on desktops and costs $1,000 per user for a standard edition and $1,600 per user for a professional version. Will this company be successful and become another Google? Read more and decide after looking at an example of database drilling."
Seems like every tenth article is there to provide a link to Piquepille (or however you spell that asshat's name) and his blog. Why can't he just write a long story submission, and the editors display the first paragraph of it?
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Just so you people know, Roland Piquepaille (the submitter of this story) has a growing repuation as a "blog spammer". That is, he sends in stories to slashdot compulsively (and I assume sometimes repetitively to get it on the front page) which always include a link to his blog at the end which provides him revenue from the ads on his site.
I'm not going to go as far as a lot of people who post about this and claim that this makes him an inherently evil force that must be stopped, it doesn't, but I'd just like people to be aware of this. I mean, his blog entry on the topic is usually just a rehashing of the articles submitted adding nothing. I really think the editors should edit out the compulsive blog link, but whatever, there's a lot of things we all think the editors should do that they don't.
Let me just give you the one feature which I think makes this extremely useful:
Don't get me wrong. I'm a CLI type of guy, but the truth is that we live in a graphical world, and I get paid to provide users what they need to make their jobs easier.. I'm pretty sure this will help them.
---
There are no data that cannot be plotted on a straight line if the axis are chosen correctly.
Unstructured data? What are you talking about? Data is by definition structured!
This is a common term in the database, search and information retrieval fields. Broadly, "Structured data" refers to information that is split up into well-defined component fields; "unstructured data" is data in one undifferentiated field.
As usual this is context-specific and not truly a binary distinction, but consider an HTML web page that has been generated from a database. In the database the information is highly structured: stored as fields that have both syntactic and semantic rules associated with them. On the web page you have essentially a block of text, usually with minimal structure to it. Both contain the same information but one has lots of structure, the other has much less.
SQL is a good language for querying structured data, Google is a good "language" for querying unstructured data.
Sailing over the event horizon
This is crap. They only interface with MS DBs.
FTFA:
Specifications: The Standard Edition of Tableau connects as a client to three types of databases: Microsoft Access, Microsoft Excel and text files. Tableau is not a "data silo." Rather, it issues queries to these existing data sources using standard drivers.
Requirements: Windows 2000 or later release. 30MB hard disk memory
took into account the links _to_ a page as a measure of it's relevence.
The idea being that the more linked to a page is, the more value it has - thereby using people as a way of meauring the worth of a page. By examing the words people link with, as well as allowing Googlebombing, it sidesteps meta-tag pollution etc.
Been de-emphasised, compared to other sub-algorithms, but it's not just the appearence that set google apart in the early days. Before they had ad's.
"Early days" *shiver* I can remeber when AltaVista was the pinnicle of web searching, and using Archie and Veronica.
To be fair, English is not his first language. He means motivation (animation in french can mean this). Maybe next time he won't just shove his french CV through Babelfish.
However consider that the content in a big text field still has structure. If it is text data it is comprised of paragraphs, words, sentences, letters, etc. -- the structure is there just slightly more difficult for computers to work out.
Does 'common usage' trump the 'actual' definition here (e.g. structured vs. unstructured)?
I wish it didn't. Personally, as one in the DBMS field, I would much rather prefer people not use unstructured incorrectly (as 'common usage' does): technically "unstructured data" is an oxymoron. Data has structure otherwise it is not data (just random noise?).
Thanks,
--
Matt
This is the Stanford spinoff of the Polaris project presented at IEEE InfoVis several times over the past few years. Chris Stolte was the main student involved, and a CiteSeer search on his name will turn up most of the related work. To summarize: The goal of the work was to provide a visual programming environment (using a spreadsheet-like layout) to presenting data in multi-dimensional databases. It uses some sound (a.k.a., proven) results to create initial intiutive mappings of this data. See the papers for more details.
It depends entirely on the context. Unstructured data is an oxymoron in information theory. Calling it an "incoherent term" is essentially correct. If there is no structure, there is no information. If there is information, there must be structure.
In the database world, on the other hand, it essentially means that the structure is something other than the record/field structure used by databases.
Data mining is sort of in between, and "unstructured" there just means that the particular analysis tools you're using can't parse it, not that it couldn't be parsed.
If I write, "three point one four one five ", your data mining tool would probably call it unstructured, but if you know that the next characters will be "nine", you have proven that it is not unstructured at all.
So the database and data mining terms are weak, though meaningful in the context of their own tools. The information theory definition is the most fundamental in that it depends only on the data itself.
"Those who have never entered upon scientific pursuits know not a tithe of the poetry by which they are surrounded."
This post caught my eye both because, I use Tableau everyday and because without Google, I never would of found it. I am a principle developement engineer at a large manufacturing company. I am often asked to solve problems that require determining the relationship and relative importance of dozens of design choices and or to tune a dozen or more parameters to optimize the performance of the system. While I am a good programmer, I am not a database expert. Prior to using Tableau I was using Excel pivottables or custom programs to explore the relationships amoung variables. This is a tedeous and time consuming process. I have also used Matlab, IGOR, and Origin where appropriate. For a recent project the amount of data overwelmed my usual approaches. Not having time to reinvent the wheel I went looking on Google to find someone that had already solved the problem of exploring unknown relationships in large masses of multidimensional data. I quickly found powerpoint presentations describing Stanford's "Polaris" data visualization tool. The website stated that a comercial version of the tool was in development. I then emailed the project lead, Pat Hanrahan, http://graphics.stanford.edu/~hanrahan/ As he was on vacation, it took over a month to get a response. He pointed me to his start-up Tableau-software. They were just in the process of releasing a beta version. I fell in love with it immediately. I could for the first time, get an intuitive understanding as to how all the variables in the system were interacting. And in just minutes after collecting the raw data. Tableau can look at datasets in the millions of records. Excel is still limited to 2^16-1 records. Why? Could Microsoft adopt many of the features of Tableau? Sure. Will they? Not likely. Excel, after nearly 20 years of development, still does not have a decent default plot format that is acceptable to a scientist or is near publication quality. Instead they look like poorly drawn cartoons. So you are forced to use much more expensive programs like Matlab, Origin, or Mathematica to get decent Plots. Pivot charts are just that, charts. Not a real scatter plot of the data. I have seen engineers completely misinterpret data because they apply a trendline to a pivot chart not relizing that what looks like the x-axis is just a set of labels and the x-axis Microsoft accually uses in the trendline calculation are just ordinal numbers. True engineering data visualization is simply not the market Microsoft is after. The examples displayed on Tableaus website are somewhat trivial compared to what the software is capable of. The true power of the software is the ability to explore many dimensions of the dataset all at once. Edward Tufte stated: "Graphics are at their best when they represents very dense and rich datasets." This is easy to achieve using Tableau. The company is very open to suggestions for improvements in the interface and in the feature set. They implemented in the first release alot of functionality I requested during the beta test phase. I have recieved two upgrades in the first year and an expecting another in the near furture. The President of the company visited and we had a great talk. I would love to give him example plots of my data for their web-site but I cannot as all the data is propietary. Is Tableau software the next Google. Of course not. That is simply a stupid, idiotic statement obviously made to draw attention to the post. But don't let that detract from what is a great piece of software if you have the right problems to solve.
Beer for Dolphins?
-----
PGP Key ID 0xCB8FF658