 |
.. (לתיקייה המכילה) | |
Should we filter the amount of followers we print to include only those we scan in the graph?
|
No. The amount of follower should be the real number of followers, as stated by Twitter. |
We noticed that we need to implement the ability to say "stop if you visited L nodes" but in the description of the Mains, there is no mention of it. Should we not limit the run with this value?
You should. While we indeed forgot to mention this in the main description of the assignment (corrected), you should note that the build.xml files do indeed contains the l argument.
|
When running BFS one of the users restricts access to its friends list.
Our JSON parser throws an IOException over the request, with the message "Server returned HTTP response code: 401 for URL: ...".
Handling that 401 error actually opens a pandora box - as there are many other possible error responses from the server.
So - are we required to distinguish such errors from other IOExceptions such as timeouts and continue crawling?
If so - should we count the user as visited (the crawler implementation we use first performs the action on the node and afterwards expands it).
|
Such cases will not be checked in the automatic check. To be on the safe side - you should log these errors on error or fatal levels. If you decided you can continue the crawling (i.e. used error level) you should count these nodes as visited. |
In the assignment it is written that we should make only two API requests per user.
How does this correlated to the fact that we need to find out 3 things:
1. The user screen name.
2. The list of followed users.
3. Number of followers of the user.
|
All the information you need can be found in the returned answer of the two API requests mentioned in the Assignment. Look at the data that is returned, and you'll find everything you need. |
When a user follows more then 5000 users the returned answer is broken into 5000 people segments by the API.
What should we do in such cases?
|
We will not check you program using users that follow more then 5000 people. To be on the safe side, you should only make one query for the followed users list (per graph node), thus you should work only on the initial answer of 5000 users. The example given in the Assignments section was given using the TechnionLive Twitter account that is following 30,000+ users. A second example will be published shortly. |
Should we limit the BFS run by the number of nodes L?
|
Yes. As you can see from the build.xml file you need to pass the L parameter for this main as well. |