We're developing a website that will try to have a lot of interesting quality content and hopefully will attract a big, vibrant community. So, we expect that our future users will comment a lot.
Time and resources for rolling our own solution are not an issue in this stage, but I'm wondering if in the long term this is the best idea, considering the trends.
The way I see it there a few solutions and I'd greatly appreciate it if you could chip in some advice on choosing the best or suggesting an entirely new one:
Rolling our own solution, tied to our member system (until now this
is what we've used, it feels behind the times somehow + it might be hard to keep up
feature-wise in the long run with something dedicated like Disqus)
Our own solution + Login with Facebook, Google,
Yahoo etc (more convenient for users, but I feel that we're losing
the user base)
Using something like Disqus or Facebook comments
(convenient, feature rich and very easy to maintain, but total loss
of control and questionable SEO, even though Google indexes Facebook
comments and recently content generated by JS)
+1 for your conviction that user comments, even (or especially) negative comments, are crucial to your success. Bravo! to you.
Of course, an ideal forum is the one we're using now. There is a process advertised for making an SE-type forum for your site. It depends on generating enough interest right here on SE, but maybe it's worth a shot.
I have wondered for long if a WP-style blog could succeed as a forum-like vehicle if blog comments were not merely enabled but encouraged. Have you thought at all along those lines? I'd like to hear your thoughts if so. To me, Facebook is a giant pita: I hate the sucker. But I'm an FB noob and you've no doubt had more success than I in rasslin' that sumbitch into submission.
Edit: As to getting revisits: you notice that SE does this by "gamifying" the forum: awarding reputation and badges. Good, thoughtful post on the subject at Coding Horror
Related
Can anyone recommend a good option to translate websites into Spanish? We tried using the Google translate plugin but the translation was so rough (very inaccurate, bordering on embarrassing the company) we had to hire a company to refine the translation so that it was much more accurate which makes for an extremely inefficient process for updating the site moving forward.
We're in health insurance, so the language we're translating is very specialized in nature and needs to be accurate for our members. To make it even more complicated, the Google Translate plugin happens instantly, so the translation is live before we have a chance to refine it before users can see it. In other words, there's no way to refine the translation before you make the content visible to users in the production environment. This is a legal regulatory requirement for Covered California and the Affordable Care Act, so it has to be a top notch implementation.
Short of a proxy solution that intercepts the content before it hits the production site or a separate site coded in Spanish, I'm not sure what other solutions exist if any. Ideas? The separate site solution is also problematic because it requires a bilingual staff and it doubles the work because both environments have to mirror each other exactly at all times.
Recommendations? Ideas? Any suggestions based on experience are most welcome!
Hire developer - he will describe all you need. You will never do it by your own. If you already have - hire new one, he will know how to do it. Question is very spiciefied but any (let's take for example php) php-engine (framework) or even custom php-engine can be updated the way you want.
Preview before upload to public? Easy! Change by moderator|admin values of translations? Easy! Main thing that each sentence (or even paragraph) you will describe by your own... I don't want describe all mechanism of it - hire developer and he will do all you need. $)
I am looking for algorithms that allow text extraction from websites. I do not mean "strip html", or any of the hundreds of libraries that allow this.
So for example for a news article I would like to identify the heading and all the text, but not the comments section and so on.
Are there any algorithms for that out there? Thank you!
In computer science literature this problem is usually referred to as the page segmentation or boiler plate detection problem. See the report Boilerplate Detection using Shallow Text Features and its related blog post. Also, I have a few reports and software sites bookmarked that address the problem. Also, see this stackoverflow question.
there are a few open source tools available that do similar article extraction tasks.
https://github.com/jiminoc/goose which was open source by Gravity.com
It has info on the wiki as well as the source you can view. There are dozens of unit tests that show the text extracted from various articles.
"Content extraction" is a very difficult topic. There are no common standards to identify the "main-article" content (there are several approaches to make HTML easier readably for crawlers, e.g. schema.org, but none of these is very popularly used).
So it turns out, that if you want good results, its probably best to define your own XPath selectors for each (news) website you want to scrape. Although there are some APIs for HTML content extraction, but as I said its very hard to develop an algorithm which works for every site.
Some APIs you could use:
alchemyapi.com
diffbot.com
boilerpipe-web.appspot.com
aylien.com
textracto.com
What you're trying to do is called "content extraction". It turns out to be a surprisingly hard problem to solve well, and many naive solutions do quite badly.
Instapaper and Readability both have to solve this, and you may learn something from looking at their solutions. They also both provide services that you may be able to take advantage of - perhaps you can outsource your problem to them and let their API take care of it. :)
Failing that, a search for "html content extraction" returns a great deal of useful results, including a number of papers on the subject.
I compared a few different libraries, and had really great luck with Mozilla's Readability library (Node), or its Python wrapper.
For example, take this CNN article: https://edition.cnn.com/2022/06/01/tech/elon-musk-tesla-ends-work-from-home/index.html
Readability successfully returns only the relevant data:
New York (CNN Business) Elon Musk is demanding that Tesla office workers return to in-person work or leave the company. The policy, disclosed in leaked emails Musk sent to Tesla's executive staff Tuesday, was first reported by electric vehicle news site Electrek. "Anyone who wishes to do remote work must be in the office for a minimum (and I mean *minimum*) of 40 hours per week or depart Tesla. This is less than we ask of factory workers," Musk wrote, adding that the office must be the employee's primary workplace where the other workers they regularly interact with are based — "not a remote branch office unrelated to the job duties." Musk said he would personally review any request for exemption from the policy, but that for the most part, "If you don't show up, we will assume you have resigned."
etc.
I think your best shoot is study what information can you get from the metadata and write a good html parser, oEmbed could be a good standard =)
https://oembed.com/#section7
I'm in the process of developling various applications for whom the end users are both engineers and salesman. Some of the operations and options may not be immediately obvious to all users. All applications are delivered with a PDF and paper manual - but of course nobody reads them!
I would like to improve the usability of the applications by including dynamic context sensitive help. One option would be alá MSDN and have F1 call up a web page - however internet access will not always be available and even this will be too much effort for some.
Another idea is to have descriptions pop up when an option is hovered over - like a tooltip.
I'm interested in other peoples views on this and what are best practices in this situation. Along a similar theme to this post What are common UI misconceptions and annoyances? I'd like to start a discussion regarding these two points:
What would be the best way to go about it?
What help features in existing applications you use either delight or annoy you..?
In my experience nobody but programmers reads the help. So when you have a technical and non-technical target audience you end up providing 2 ways of doing everything:
A Wizard with a few options.
A property editor with lots of options.
In either case, pictures are usually better than words for documentation. So a screenshot or 3 with big green arrows and circles calling out what does what will go a lot further than an indexing, exhaustive help file.
In my experience it would be very helpful to have a tooltip on each option that provides a little more definition/clarity for each option. Additionally, you can improve usability by having the default screen contain a few common, simple options and providing an advanced section that provides more control.
I'm currently working on a similar side-project. We have an existing product that's used by people as part of their day job. There is an inherent learning curve on the product, so users receive some degree of training and have people they can turn to for assistance. Even so, we know it needs more help and user documentation in general.
We are starting this help enhancement project by running a quick survey on the end users, (offering a prize draw as an incentive). We will also speak to the support staff who have to deal with help requests. This will uncover some of the pain points, and will give us a clear idea of how to focus our time & resources.
Guidelines on when to use inline tips vs tool tips etc can be found in various style guides, e.g. here:
http://developers.sun.com/docs/web-app-guidelines/uispec4_0/11-help.htm
Bear in mind that it's probably a bad idea to just copy & paste the text from your existing manuals into contextual help tips. You're going to need help writing completely new content. See if you can get some time from a technical writer / copywriter.
I definitely appreciate a good interface and as a developer, I try to create them for my users. But appreciating a good interface and designing one are a different thing. I'm looking for good interfaces (such as IMHO StackOverflow, Gmail) as examples of good UI from which I can model my own UI's.
I personally think that Netflix has an excellent web UI. Responsive, easy to navigate. Not mutch CRUD going on, but I find it very comfortable.
Pretty much anything by google, really. They're all very simple and to the point, focusing on usability.
You should get yourself a copy of both Don't Make Me Think and The Non-Designer's Design Book for your base knowledge/insight.
From there, it's much easier for you to dissect and analyze the layouts you already know and like, and recreate them for your own amusement.
edit: To mitigate misunderstanding, the point I'm trying to make is that you probably don't need as many good examples of nice layouts, if you know what to look for. For example, I can be shown a thousand haute couture dresses, and I still couldn't make one myself, because I don't know what to look for.
My favorites
Stack Overflow: This is a WIKI so it's not a rep point grab. I just really love the interface on this site. Been to too many crappy Q/A sites
Google Reader
MSDN: It's gotten a ton better in recent years and is a great way to grab little esoteric details about various APIs
iStockPhoto.com it's simple, effective and handles a large amount of information and data without getting bogged down. It also doesn't get in the way of the info you are looking for.
A good user interface fulfills a specific need of its users effectively.
As an example, here is a site (translation) that I have created for finding out what food is available in the cafeterias of the University of Helsinki. The typical use case is that when a student is hungry, he needs to know what food is available in the neighborhood student cafeterias (which are cheap for students), so that he can choose where to eat and what. He knows where each of those cafeterias is, but does not know what food they have today.
That site shows all the needed information at once. Because the students typically have a couple of cafeterias where they go, they can either bookmark the page with those cafeterias selected, or save the selection as a cookie. After that they can reach their goal without any navigation on the web site.
I don't use it on a day-to-day basis, but I'm very impressed with the Perseus Project digital library.
Here's a link to a poem from Catullus' Carmina in Latin as an example of the interface. Some features that I really like:
Click on the bar near the top to jump to any poem in the work. Larger chunks of the bar represent larger sections of the work (poems, chapters, however that particular work is logically broken up by the author).
Click on a Latin word in the poem to bring up a window (be patient; it seems to take a while) with lexicon entries, user voting and statistics on the word form (i.e. what the inflection means in the context of the sentence; it can be ambiguous in Latin) and so forth.
There are a number of resources down the right column, including various English translations, notes, references, etc. Any of them can be either shown in the right column, or swapped out with whatever is in the main content area in the center.
One of my personal favs: newspond.com
I remember reading a how-to book several years ago, about Use Cases. (This was probably before user stories supplanted this part of the terminology.)
The task at hand was something like adding new customers.
There was roughly a 1-page offset section that described a couple developers who said something like "We don't need no stinking Use Cases. We do this all the time. Won't take more than a couple hours."
Next day they came back with your classic CRUD-type table maintenance screen - and were consequently chewed up in comparison to the UI developed from a good Use Case.
Anybody recogize this? It made a big impression at the time, and it's still a good cautionary tale. I'd like to find it again.
I am not certain, but it sounds vaguely like something from The Inmates Are Running the Asylum by Alan Cooper.