Microsoft Research working on making Excel easier to use

Microsoft has made a number of improvements to its popular Excel spreadsheet program for its newest version, Excel 2013. However, it can still be something of a pain to use the software for various tasks. One of them is labeling which parts of an Excel document are in different categories

Normally, users would have to manually type in those categories for each entry on the spreadsheet. However, a new project from Microsoft Research is working to automate that kind of task via machine learning. In a newly posted video on the division's YouTube channel, researcher Lucas Bordeaux shows their system at work.

Bordeaux shows a typical Excel expense spreadsheet and then types in a category inside one of the entries, "food". The Excel embedded software created by Microsoft Research then scans through the spreadsheet and figures out which of the other items in the expense report can also be put into the "food" category.

Another example in the video shows how the embedded program can be used to label election results if the spreadsheet doesn't have complete labels for all of the candidates in the election but selecting the entire spreadsheet and then selecting an "auto complete" button which fills in the blanks.

So when will this nifty feature be put into Excel? That's a good question and one which the Microsoft Research article doesn't really answer. Too bad. We would love to have it ready for the next update for Excel 2013.

Source: Microsoft Research on YouTube

Report a problem with article
Previous Story

Chinese PC maker showing a Surface Pro knockoff at CeBIT 2013

Next Story

Microsoft Research shows Kinect can be good with hands; example of Kinect 2.0?


Commenting is disabled on this article.

It actually looks as if that is based on an internal build that is on a different branch from Excel 2013 RTM, portions of the UI still look like 2010.

I would guess there are different versions all over the place that different teams are using on different research projects.

Most of the time I use Excel, someone always wants a column header changed to something custom anyhow.

Besides, it could only determine some of the columns. For instance, say you had a column with a company name, then 10 columns which contain only yes/no values. You'd still have to label the 10 columns anyways as it has no clue to what the data refers.

Hmm, I'm not sure if I could trust machine learning to determine metadata, especially if I'm not personally checking through what it's outputting (such as a long list of data).

I hope they don't change too much and keep the classic functionality, i know ALOT of people especially in the financial sector that basically live in Excel, which i don't blame them, excel is near perfect for a lot of numeric working, the add in in the newer versions allowing massive data manipulation with data sources such as SQL Server was the icing on the cake!

When it comes to easily relating data for custom reports across any number of separate SQL resources, I live and breathe Excel, though I'm a little curious about playing with SharePoint and seeing if I can build dashboards the same way.

It basically means I'll finally be able to win the big jackpot just by pasting winning number history into this amazing Mïkrosófte Excél thing, right?