Information Retrieval and Web Search COMP6714 24T3

Questions after the exam

Posted by Raymond Wong Monday 02 December 2024, 02:28:09 PM.

We have started receiving email inquiries regarding exam marking, course performance and special considerations right after the exam. After replying to a few, I have decided to summarize the rest of the inquiries here (and will not reply individually) so that we can focus on marking the exam and reconsidering some of your projects:

1. a) Discrepancies in the last two digits of the required decimals for Q1, and +/- 1 of the last digit of the two decimals in the short answers will be ignored. b) There are no partial marks for multiple-choice questions, but there are partial marks for answers close to the expected answers. c) We won't respond to how a particular question is marked, but we will adhere to the marking criteria and ensure consistent marking for all students.
2. Some of you failed to submit the files. If your work has been saved, we will try to check your saved files during marking.
3. Some of you forgot to save any changes to the files and have requested us to mark your results from the scribble paper instead. Unfortunately, as with other UNSW exams, we are not allowed to mark students' scribble papers. You were also explicitly warned to save your work before submitting in the last lecture, and the exam instructions have a bold statement "SAVE your changes and SUBMIT each question."
4. Your project will be manually inspected, and results may be scaled up at our discretion after we finalize your exam marks if you perform well in the exam but not in the project. This will apply to everyone, so there is no need to send us an email to apply.
5. If you are worried about the exam, let's not worry about the marks for now. Since we have just started marking and have yet to scale, your concern might not even be necessary. We will try to scale as much as we can, to the extent that is fair to others, by considering your assignment and project performance.
6. For those who have sent us other personal reasons, unfortunately, we can only assess and mark based on your academic performance on the course assessments (while I may personally sympathise your situations). We can scale the individual components, but only to the extent that is still fair to others in the class. Any special considerations will have to be submitted via UNSW Special Consideration: https://www.student.unsw.edu.au/special-consideration

Project results

Posted by Raymond Wong Monday 25 November 2024, 12:26:40 PM.

Your project result has been sent to your UNSW email.

1. If you have questions about the marking tests*, please post them in the Ed forum.
2. If you have issues about your marks (i.e., you believe there are discrepancies after checking the marking tests), please come to the consultations this week (check the timetable on WebCMS).

Please note that email inquiries about marks (e.g., why you received low marks even if you passed the sanity test) are likely to receive slow or no responses during this busy time of the term.

The same marking tests, developed based on the provided specifications, are used across all submissions. If you request us to consider different marking criteria for your particular situation, we would have to decline as that would be unfair to other students.

As mentioned in the last lecture, your project will be manually inspected, and results may be scaled up at our discretion if you perform well in the exam but not in the project.

*You can check the actual marking tests by running ~cs6714/reuters/marking on a CSE Linux machine, as previously posted.

FYI, the cohort performance of the project is as follows:

 1st Qu.: 6.140  
 Median : 7.930  
 Mean   : 7.773  
 3rd Qu.: 9.485  
 Max.   :11.000  

 17% of submissions achieved 10-11 out of 10

Project marking tests

Posted by Raymond Wong Thursday 21 November 2024, 10:57:08 PM.

As mentioned in the last live lecture, we are finalizing the project marks (almost there) and running the tests a few times for cross validation. Your marking result shall then be emailed to you in 2-3 days. Once it's done, I will post another Notice here.

Meanwhile, if you like, you can check and run the marking testcases by:

1) log in to a CSE linux machine and go to the folder containing your index.py and search.py

2) run ~cs6714/reuters/marking

In theory, the results returned from this script that you run should be consistent with the marking result that you will receive.