Tag Archives: algorithm

Apr 11 2012

Twitter Just Got A Lot More Interesting, with Prismatic

 

Even after four years of tweeting, Twitter is still an enigma to me. Mostly because I find myself constantly flooded with information from it and I am unhappy with my inability to segregate the really good stuff from the really useless stuff. I prune my follows regularly and use all sorts of apps to bring the best stuff to the top, but I still am plagued with the idea that I am missing real gems in the process. Don’t get me wrong. I am not saying Twitter isn’t useful – I definitely see it as a valid news source, particularly for real-time and breaking information. It’s just that my efficiency-first personality renders me constantly frustrated by the process of reading and rating the worth of tweets.

 

Speaking of gems, I may have found an answer, at at least, a better answer to may age-old problem of information overload in my Twitter stream. It’s called Prismatic. Prismatic is an algorithm driven application that reads through the Twitter stream, and reports back with the tweets and links that will likely be of the greatest interest to you. It’s awesome for a number of reasons. First, while you may have the coolest Twitter follows in the world, they are only human and may miss some key information. Furthermore, if you have narrowed your interest graph to a particular topic or set of topics, what are you to do when you need information outside that sphere. Prismatic doesn’t just look at your Twitter follows, it looks at the WHOLE Twitter stream (yikes!) and then promotes the best stuff on your area of interest, while still offering a series of links along your sidebar to permit you to investigate different topics and tags.

 

Prismatic’s developers bill it as a discovery engine and have leveraged their deep knowledge of machine learning and linguistic programming in building the app. So what does it look like? After answering a few questions to get your page started, you will be greeted with a clean, web-news page look, with main stories in the larger left side, and a sidebar along the right. Hover over the home icon on the right to see your “home” feed, which is built around your interests. Hover over the globe icon on the right to explore the larger universe of available topics. It appears that the Global feed is also personalized to you, but covers a broader range of material. You get suggested topics as well as suggested publishers, so you can lock in on the most interesting stuff. You can also employ the search box on the right to get right to what you are looking for.

 

Of course, you can fully interact with the articles presented to you in your feed. A typical blurb looks like the image below, and  you get infinite stories by simply scrolling down the left side of the page:

 

 

You can see the article at the link, an image and a bit of the text. Click on it to go to the original source article. See a few of the tweets that have linked the article to the right. Click the “x” if you aren’t interested and would like it removed, click the “+” if you want to favorite the story for later reading and click the comment bubble if you want to share the story elsewhere. You can share by email, Twitter or Facebook. You will also see how many people have shared the story, so you can get a sense of how popular it is. Click the eye at the top of the page to show that you have read all of the articles. See and click on the tags that brought the story to you, and click on the related stories link to see all stories Prismatic has related to the original story. This makes it easier to dig deeper into a particular bit of news.

 

I am loving this app. I found in the first few articles of my feed some great stuff that hadn’t percolated its way up to the top in Feedly or my other “smart” aggregators. I was able to quickly read and tweet it back out. Very efficient. If you are having difficulty managing the wild Twitter beast, then consider trying out Prismatic. My one complaint, lack of a mobile app, is being addressed as we speak and hopefully will be out shortly. Prismatic on the iPad would be an absolutely killer combination.

 

The service is free, but it is still in limited beta, so you will need to request an invitation at their website. It didn’t take long to get mine. Head on over, then head back here with your feedback on this great, intelligent Twitter-taming tool.

Advertisement
Oct 14 2010

A Little Late To The WestLawNext Party

But, better late than never, so they say. After boning up on everything WLN hot on the heels of the rumors and ultimate announcement of the revolutionary new legal database search interface earlier this year, I came away with the impression of “cool, but not worth the extra change.”

Just the other day, as I was hopping onto Westlaw for my daily visit, I noticed a little orange link at the top with admonishment that, for a limited time, I could try WLN free of charge. I am not one to turn up any chance to play with a new toy, particularly a free chance, so I decided to run my rather arcane inquiry in WLN instead of the old interface.

I am not going to do a full blown review of WLN here – there are scores of great posts and articles out there that lovingly list out every feature and improvement.

All I want to say is this: that new search algorithm West has outfitted WLN with really does improve your results. My query was on a very fine point of insurance law – I was having some difficulty even understanding the question, let alone formulating a tight search for an answer (after this many years in the biz, it takes something else entirely to make me scratch my head). So I entered my mostly unformed inquiry into the search box and, to my surprise, the very first hit was directly on point. I can only imagine what WLN would do with one of my familiar searches.

The bottom line question for me on the issue of WLN has always been: is it worth the money? Before I would have responded, unequivocally, no way, Jose, I can get what I need just fine from the old interface. Now, I must qualify with the further response that, if you have matters requiring turbo-charged research in unfamiliar waters or cases where the stakes are higher and mistakes more expensive, then WLN may be a reasonable cost of doing business.

Aug 31 2010

Get Ready for More Organized Gmail

Google’s at it again. This time, the object of its affections is Gmail. Today, Google will be rolling out Priority Inbox (link here). In a nutshell, application of this feature will automatically sift to the top of your email pile the messages deemed of greatest importance to you. When enabled, all of your messages will be sorted into three groups – “important and unread”, “starred” and “everything else” – and will be slotted in your inbox accordingly.

Look for the “New! Priority Inbox” message on the top right of your screen (it should be there within the next week or so). Then, simply set it up in your account and click yes when it asks you whether you want to enable the system. There are a few set-up options, such as manual selection of priority contacts and order of inboxes, but most of the magic comes from Google’s ability to read how you read your email. Google looks at how the mail is addressed (whether just you or to a group), which emails you open and respond to (and how frequently you do so) with respect to a particular sender and keywords in the emails you most frequently read. As you use the filter, you can instruct Google when it makes a mistake, such as starring messages that Google thought were unimportant and demoting those Google wrongly pushed to the top. Over time, it will learn your preferences.

Reports from people who have been using this service before the public roll-out indicate that the service is very good, but not perfect. But perhaps they haven’t spent enough time yet teaching the algorithm the ropes.

Obviously, from the email sender point of view, it will be important to consider how to phrase your message so that someone else’s Priority Inbox doesn’t demote your missive to the bottom of the heap. In other words, consider when and how to use group email addressing and what keywords to include in your message to ensure it will rise to the top. Sort of like search engine optimization for email.

Looking forward to trying this out – I get a lot of dreck along with the diamonds in my Gmail inboxes – interested in seeing just how effective Priority Inbox can be.

On a humorous note, I learned that Google considers email that is a cut above “spam” but lower on the priority totem pole to be “bologna” or “bacn“, the latter being email that a person has subscribed to, but isn’t regularly read or responded to. I thought people on the internet thought bacon was a good thing. There I go, learning something new every day again.

Jul 1 2010

Lock Down Your Email

If you suddenly find yourself needing secure email, don’t worry – head over to Encrypt Easy (link here), a software purveyor that offers a free web service for encrypting email using the Blowfish algorithm. It couldn’t be easier: select a password for your email, type your message into the Web editor, then select “encrypt”, “decrypt” or “clear.” Give out the password to the desired recipient, and then save your message, email your message or even use it as an instant message.  

While the stripped down web service is free, it supports Encrypt Easy’s paid product: encryption software that protects private data, including files, folders and entire directory trees. The program features fast and simple one-click encryption using secure algorithms, as well as shredder and wipe-out utilities. The beauty of the software (built into the Web encrypter as well) is the fact it creates a self encrypting file, making it usable for sharing secure data with others who may not have their own encryption / decryption software.

Want to know a little more about strong passwords, the best encryption algorithms, or the whys and hows of encrypting? Check out their article links here.

May 11 2010

Genieo – Another Means for Auto-Curating The News

I have talked about intelligent news readers here in the Studio before, so why not again? Genieo (link here) is another entrant in the arena of relevance-based reporting. Genieo is a desktop application based on proprietary algorithms that “read” user behavior, learn relevancies and feed back content tuned to user interest. That information is viewable through a dynamic, personal content portal, a/k/a home page. From the site:

The company’s vision is to become the leading source of user behavioral modeling, supplying content and application providers with effective means to address their users’ needs, and help advertisers and publishers with targeting personalized ads, with the utmost precision.

The process is simple enough – download the Genieo software and make your settings, allow Genieo some time to pull together your interests and profile and create your home page, and open your browser to your page. It will show, in magazine format, top news headlines, actual and real-time updates and filtered Facebook and Twitter updates. Everything shown will be related to your interests and Genieo will refine those interests over time as you use their service. You can also manually adjust interests to direct the process.

Some cool adds? Genieo will auto-update your bookmarks as you visit sites and manage these bookmarks based on your level of interest and interaction with them, all without interfering with your browser’s own bookmarking system. A Mini Topic Filtering System sifts through all information at the “highest resolution”, continually bringing items exclusively relevant to you. You can pull real-time updates from events you are following, stocks, sports (WORLD CUP!!!!) or developing news stories. Unobtrusive notifications of updates are displayed on your screen as you work. And, of course, there is the ubiquitous ability to one-click share your stories with your networks.

Genieo offers an interface modded for iPhone – simply navigate to my.genieo.com, log in, and get your information iPhone-sized. You can also pull your Genieo feed into your favorite RSS reader.

Lots of relevancy-based coolness from the fine folks at Genieo! Curate your own news now, automatically!

Apr 27 2010

Grasping For Meaning In 140 Characters

Seems anything can be the subject of scientific study. “Augmented social cognition” researchers at the Palo Alto Research Center are groping for automated ways to cull the “meaning” behind tweets, per MIT’s Technology Review (link here).

The problem: users want to be able to scan a timeline to get the gist of what is going on without having to read every single tweet. This is a real problem when you have more than a 100 follows.

A portion of the study has been devoted to developing a recommendation system, not dissimilar to my6sense, which determines which tweets are most relevant to a user based on that user’s interaction with various tweets and with other users.

Borrowing imagery from a real-world stream and it’s eddies, the Eddie project is also working a “topic browser”, a machine aided inquiry into a Twitter stream that should enable the viewer to get more than simply the keywords; to reach the actual “gist” of the stream.

The topic search aspect is more problematic to implement in that natural language searches apparently rely generally on more text than the 140 character space limit allows. Additionally, there is simply such a vast amount information that it becomes logistically difficult to parse.

The method the research team has developed for dealing with these challenges is to treat a tweet like a search query. Tweets are filtered by removing unnecessary terms (like “RT”). Then the significant terms are pulled using the algorithm, and then run through a search engine, in this case Yahoo’s Build your Own Search Service interface.

In essence, tweets are matched with search results, allowing for a “best guess” as to what the tweet means. Given, however, that the major search engines are now indexing tweets, there is a real possibility that the tweet topic browser could return search results that mirror the original tweet.

The researchers anticipate that the topic browser may be online for live testing this summer. I look forward to playing around with this interesting combination of tweets and search in the pursuit of Twitter-meaning.

Hat tip to Resource Shelf