That's what is says at the top of Paolo Massa's blog: "Ramblings on Trust, Recommender Systems, Social Software, and much more." Wow, isn't that a great combination? Personally, I believe that the combination of these things has the potential to become a major new direction for the social web and the new social marketplace.
With entries like Soft Trust and Hard Trust and Evaluating Collaborative Filtering Recommender Systems, Paolo discusses these foundations of the social web of the future.
He's not just rambling about these things, however, he is also studying them at the PHD level. An example is a PHD research proposal titled Trust-Aware Decentralized Recommender Systems. Admittedly, I haven't read the 20-page propsal yet, but with a title like that, I certainly plan to. It will probably end up as the topic of a future Web Dawn entry.
Personally, I need to make some time to read more about these topics, and Paolo's blog is near the top of that reading list.
I recently finished reading Six Degrees: The Science of a Connected Age by Duncan Watts. While I was expecting to read a lot about social networking and the degrees of seperation in social society, I suprised to find so much science. Yes, I know: the subtitle should have given it away. But this is not a problem at all: the book is written very well, in such a way that anyone can understand and enjoy. In the fact, this convergence of science and sociolgy is makes makes this book truly interesting. The book discussed various types of networks, with results that are surprisingly similar. I recommend this book to anyone interested in social networking. The nineties were about the web network, and the decades that come will be about social networks that manifest them themselves on- and offline.
I am developing a new site for a client: Job Search Blogs. The site offer free blogs to job seekers. Job seekers can use blogs to promote their skills and experiences to prospective employers. I think this has tremendous potential for improving the recruiting process. Let's face it: conventional resumes are dull and boring, and don't really say a lot about the candidate. Job Search Blogs have the potential to act an an extension of the resume, providing insight into the thinking and personality of the job seeker.
A few months ago, when I pointed a friend at some of my blog sites, he replied 20 minutes later saying, "I've learned more about you in the past 20 minutes than I have since I met you". If blogs can this potential for revealing information about people who know each other, then this could be immensively helpful to an employer trying learn about a job candidate.
I am interested to learn what others think about the concept. Any and all feedback is appreciated. Thanks in advance.
Google announced an upcoming free email service yesterday - as everyone knows, as "is it an April Fool's joke" debates raged all over the Internet.
The implications of the new service could be wide-reaching. It goes well beyond a 1 GB mailbox. It's natural that this has become the focal point of conversation, in contrast to the meager 2 or 4 MB that you get with other free services. The business case, of course is Google AdWords, serving text ads targeted to the text of the email that you are reading.
Once you start to look beyond those two aspects, however, it starts to become very interesting. One day, Google announced a personalized-search demo, and now email. Well, nothing is more personal than email (spam excluded). The problem with personalization system is that you have to fill out (sometimes huge) questionnaires about yourself, which few people want to do. The most successful systems, don't require this at all. Amazon is a great example. Based on the things you search for and the items you click on, Amazon starts to recommend this that you might like to to buy - heck, they even create a "Mark's Store" tab for me. Amazon builds a profile automically by tracking your actions on the system. Now imagine that a certain company knows what you are reading about (and writing) in emails. Further assume that the same company also knows the things that you are searching for, and maybe even the news that you are reading, the products that you are shopping for, and even where you are. A much broader and detailed profile could be created by compiling all this information. This could be used to serve extremely relevant, personalized search results (along with just as relevant advertising).
Taking this one step further, we can remind ourselves that email is a form of communications. It is one medium that we use to interact with other members of our social networks. By keeping track of who we are communicating via email with, a company could build a graph or that person's social network. If the same company has detailed profiles of those people, then the potential is there to combine that data for refining search results and advertising. For example, if you email contacts think a particular search result is highly relevant, it may increase in the rankings when you search. Or if your friends or colleagues are clicking on particular advertisements, perhaps you would like to see those ads as well?
While the potential is huge, it is also a little scary. When we thinking about the collection and processing of so much personal profile data, we usually think about government spy agencies, not your friendly neighborhood search engine. Like many things relating to the future of Google, privacy could become a serious concern or barrier going forward.
Just when you thought the search engine wars couldn't get any more interesting...
Google is now one step closer to social search, with the introduction of a Google Labs demo for Google Personalized, a search service that considers your individual profile when ranking search results. This type of search was inevitable, and will go a long to improving the search results.
The next step to take into account the searcher's profile in social networking context, considering not just my own profile, but also the profiles of those whom I interact with. Ideally, social search would also take into account a social voting or scoring system, in which the search results favored by my social network increase in relevance.
The more that a search engine knows about you (including information about your social network), the better that search results can be tailored to you. But privacy issues abound - how information is too much for a company to ahve about you. There are already privacy concerns about Google and others. And we all know that Google is also an advertising company. The challenge will be to find a balance between improved search results and maintaining privacy of personal data. It will be interesting to see how this plays out.
As email spam continues to grow, the use of "whitelisting" is a growing technique. The opposite of blacklisting, whitelisting entails the creation of a list of email addresses from which you are willing to accept email. Used by itself, this technique is extremely successful in preventing spam, but it has obvious problems: 1) You have to constantly update your whitelist, and 2) you can never receive email from people who don't know you (yet). There are a number of techniques to address the latter, which usually involve sending non-whitelisted senders to a web page for validation first, which is a real pain.
Another approach is to expand the whitelist concept to include the whitelists of the people on your whitelist. In others words, allow email from friends-of-friends. And perhaps also from friends-of-friends-of-friends, depending on your personal preference. This technique relies on the trust inherent in social networks. As long a my contacts aren't friends with known spammers, it is probably safe to receive email from them also.
The system could be made more complex by adding reputation or trust-scores into the mix, which would extropolate through each degree of your social network. Email messages could then be scored according to the social network path that connects you to the sender. This approach would enable you to receive all messages, but score each based on your connectivity to the sender. Messages from senders who are not connected to you (according to your preferences) would be scored low (or high, as the case may be) and perhaps auto-sorted into a lower priority folder.
While this idea appeals to me, it seems there may be some major problems with it. First, the cumulative whitelist of my social network could be huge (millions of addresses) depending of the degrees of seperation chosen. I am not a computer scientist, but I wonder if this poses a processing problem. Also, this wouldn't work if the network data were openly and freely available. The spammer would just use the data to make sure they spoofed your best friends' email addresses every time (which would make their spam even more effective). So a system would be needed to ensure that the network information was not easily available - at least not on such a broad scale that spammers could use it to send mass mailings. In addition to these, I am sure there are other challenges. If they can be overcome, I think there is real potential in this approach as a way of filtering or scoring incoming email.
I know that I may be late to the game on this one, but I recently finished reading Smart Mobs by Howard Rheingold. Prior to reading the book, I had been thinking a lot about the potential for peer-to-peer (P2P) social networks to become the successful model of the "online" social network. Reading the book helped solidify my thoughts in this area - and expanded the scope of my thinking to other related areas and possibilities.
While there is currently a lot of talk about online web-based social networks, mobile-web-based social networks will soon become the major form of online social networks. Social networks will have both fixed and mobile interfaces, of course, and the mobile interface will become the primary one.
As I have said before, mobile or fixed, for online social networks to be truly successful, they have to be open. That is, no service provider should define or limit the scope of my social network. As it has been for thousands of years, I will continue to define my social networks - people I am socially connected to should not be required to join one - or fifteen - "social networking services". There is a role for service providers to provide interfaces and services, but my relationships should be defined independently of the service provider.
As I posted on Orkut.com about Categorizing RSS Entries:
Since this is the Semantic Social Network community, I think semantics is the answer to this one. :)
The software should be smart enough to group author-designated categories according the meanings of the words or phrases used to describe them.
I also like the way the del.icio.us social bookmarking tool enables user to specify their own tags, which others members of the (social) network can browse. Combine this with a semantics engine and an Amazon-style "the members of your FOAF network categorized this RSS feed as _________ and they also subscribed to _____________" and you have a fantasic collaborative social semantic categorization system (a CSSCS). ;)
As posted on Orkut:
P2P typically means "individuals' personal computers", but it doesn't have to mean that. For example, blogs can be viewed as peer-to-peer, making connections directly between each other using Trackback or similar technologies. I can make a trackback connection to any blog that supports tracback - on any server - I'm not limited to a particular network of sites that it controlled by a single entity.
I can also envision, using the personal computer P2P approach, a "friends caching for friends" scheme, in which willing members of your social network would "cache" your profile and other public (or semi-public) information. So if you go offline, your profile is still available.
Finally, I think that mobile devices will become the dominant "peer nodes" in such a social network. The location-based possibilities are endless. And a mobile device is more likely to be "always on".
I have started a new web site, MT Hacks (http://mt-hacks.com), where I plan to post some of the Movable Type customizations that I have done. These will include templates, scripts, ideas, and plugins.
Yesterday I posted my first MT plugin, MTDynamic. The plugin can be used to to render MT Blogs dynamically. By default, MT builds static pages - which is good for a lot of reasons. But there are cases where dynamic pages have benefits, and that's were MTDynamic comes in. The plugin enhances an "experimental" dynamic viewer that ships with Movable Type. More information can be found here.
Other MT Hacks will be posted soon.