|[UP] Content Personalization Model||Justin Scott||12/6/13 1:52 PM|
Since we announced our User Personalization initiative earlier this year, we've been talking with users and publishers and iterating on ideas of how the Web can support content personalization in which the user retains control over what is shared. Here's our latest thinking, which we'd love your feedback on.
Content websites typically cover particular topic areas; for example, a news website might know all about Politics, Technology, and Business, while a sports website might focus on Basketball, College Football, and Baseball and the teams playing those sports. Today, websites use <meta> tags to tell search engines what keywords and topics they cover in order to improve their search rankings. Similarly, a <meta> tag could be used to tell the browser that a website supports personalization and which topics it can personalize on.
When a user who has enabled personalization in Firefox visits a website that supports personalization using the tag above, the user can request more content on his or her favorite topics based on what the website supports. For example, let's say I visit the website of a fictional Outdoors/Recreation retailer called Outdoors123.com that sells equipment for Hiking, Camping, Climbing, Snow Sports, Fishing, etc. By adding a <meta> tag to its pages that lists the categories of equipment it sells, Firefox will know it supports personalization and can allow the user to request content on the topics they're interested in.
And, because Firefox has matched those topics with the user's browsing history without any data leaving the user's device, it can pre-select the topics the user is most likely to be interested in on that particular website. In the example above, Firefox would see that I have recently done a lot of research on hikes and camping, and would pre-select those topics for me to be able to request.
The user is in full control to select additional topics, de-select topics, or turn off personalization for that website entirely. And an indicator will let the user know any time a website is personalizing to his or her requests.
In the outdoors example, instead of a generic homepage that might normally feature skiing equipment, the website could focus on hiking and camping gear and show me relevant deals and top-rated products to save me time and money.
There are many specific details of how this all works and looks to the user and publisher, but to start with, I'm curious what folks think of the basic architecture of a site broadcasting personalization support and Firefox pre-selecting the best topics for that particular site.
Justin, on behalf of the UP team
|Re: [UP] Content Personalization Model||Paul C||12/9/13 4:51 PM|
As a non-expert, I have no idea how the issue of high-quality, high-relevance search results on the Internet might be resolved. None at all. All I do know is that subjectively, 99.9% of it is basically dross, often republished content from elsewhere, sometimes just plagiarised, all in an attempt to get hits via search engines, such as Google. Even with the seemingly sophisticated search filters available to Google search users, the basic process is simply fraught with frailties, short-comings and ways in which it can be exploited to the advantage of the purveyor and the disadvantage of the user, who is inevitably being exploited. When it comes to Internet content GIGO has never been truer. And the outcomes are akin to searching for the needle in a haystack. Make that an endless number of haystacks. To me, at least, there seems to be a predominance of get-rich-quick merchants attempting to get the most personal gain for the least amount of any real personal input.
Maybe one answer (amongst many others hat would be necessary) is to stand the principle of search hits on it's head by high-scoring content for it's rarity and not it's popularity or frequency.
The maxim of the majority of web site purveyors seems to be "If you can't dazzle them with diamonds, baffle them with bullshit". the bullshit taking on many forms including egregious and blatant duplication and rehashing.
|Re: [UP] Content Personalization Model||Irné Barnard||12/10/13 12:26 AM|
In agreement with Paul about the user's perceived quality of such personalization. Unfortunately FF can't do much about a site lying about its content or duplicating stuff. That tends to be handled by reporting such sites as a form of "malware". For those sites which actually tell the truth about themselves in the meta tags, and actually provide useful info depending on the meta filter it might be very nice for the user to not have to browse through irrelevant info.
It would be similar to a site like Amazon where it remembers your previous choices and starts only showing you similar products instead of the reams of things you have no interest in.
|Re: [UP] Content Personalization Model||Heather Kohlmar||2/16/14 6:51 AM|
Something that I'd personalize, which I think Mozilla should reprogram: it's minutiae, but also a glaringly bad lack of foresight.
In Firefox's HISTORY menu, RESTORE PREVIOUS SESSION really should not be directly beneath/beside CLEAR RECENT HISTORY.
I mean, really.
Hiccuping too hard while selecting the former could destroy MONTHS of work, seeing as Firefox regularly crashes, and for some reason cannot adapt to Shockwave/Flash for love nor money....
PLEASE, Firefox developers: get those two items apart from each other, and - how about a comments/suggestion box hey.
Heather Kohlmar Freedman
|Re: [UP] Content Personalization Model||Liam Bulkley||2/16/14 5:37 PM|
I've always liked the idea of a semantic Web but rdf wasnt the answer. What you're proposing looks both doable and useful even if the scope is much more limited. I'm curious how the site will be presented given ones interests. Is their a risk of this breaking the CSS?