3 ms·

We're trying to provide a better interface for searching computer science at http://scholr.ly http://scholr.ly. I'd love your feedback.

mjn 14y ago
Looks interesting! My biggest initial dislike is that the author-profile data seems extremely messy. For example, looking up my supervisor, who has a fairly unique name, his publications are split across a full 10 separate profiles: http://scholr.ly/results/?q=michael+mateas&x=0&y=0&page=0 http://scholr.ly/results/?q=michael+mateas&x=0&y=0&#... It realize it's a rather tricky problem to fix, since not every pair of identical names is actually the same person. Doing a vanity-search on myself (http://scholr.ly/results/?q=mark+j.+nelson&x=0&y=0&page=0 http://scholr.ly/results/?q=mark+j.+nelson&x=0&y=0&#...), there are 5 profiles with precisely the same way of writing the name ("Mark J. Nelson"), but only 2 of those are me. If that data were cleaner it'd really increase the usefulness of the profiles. May take quite a bit of magic, though.
- mhluongo 14y ago
  It does take some magic. Luckily there's been quite a bit of work done of the topic (eg, Lise Getoor & co http://scholr.ly/person/5486566/lise-getoor)- http://scholr.ly/person/5486566/lise-getoor)- we should show some serious improvements soon. In the meantime, I'll manually merge those profiles you've pointed out :) Thanks for taking the time!
  - mhluongo 14y ago
    Proper link: http://scholr.ly/person/5486566/lise-getoor http://scholr.ly/person/5486566/lise-getoor
tinco 14y ago
Hi, I tried a search for "model based testing" and the results seem not to be of very high quality. Are you lacking sources? Also, it's not possible to discern high-quality from low-quality papers in any way. As a graduate student, I weed out what I should read by reading the titles+abstracts in the order of the amount of citations. Citation amount maybe isn't the most optimal indicator of quality but it is a strong one. In order of importance: Scope (does it cover all journals?), Quality (how do I know if something's worth reading), Speed (I noticed your search runs a little on the slow side). Other than that the site looks great, I'd certainly switch if you would get the basics down :)
- mhluongo 14y ago
  Hm, right now we've got the same coverage as CiteSeerX (where most of our data comes from), and are working on expanding. What sources do you particularly think are lacking? We've talked quite a bit internally about whether to include citations counts. It's a tough call, but until we're happier with our scoring it might make sense to help bolster result reputation. Thanks for the feedback!
  - tinco 14y ago
    Hey no problem :) I'm just a student so I'm not researching a particular field at the moment, the Model Based Testing paper I was looking for I had to read for a course. I just wouldn't use your site as long as I feel _any_ journal is missing, since google does not give me that feeling.

mjn 14y ago

Looks interesting! My biggest initial dislike is that the author-profile data seems extremely messy. For example, looking up my supervisor, who has a fairly unique name, his publications are split across a full 10 separate profiles: http://scholr.ly/results/?q=michael+mateas&x=0&y=0&page=0 http://scholr.ly/results/?q=michael+mateas&x=0&y=0&#... It realize it's a rather tricky problem to fix, since not every pair of identical names is actually the same person. Doing a vanity-search on myself (http://scholr.ly/results/?q=mark+j.+nelson&x=0&y=0&page=0 http://scholr.ly/results/?q=mark+j.+nelson&x=0&y=0&#...), there are 5 profiles with precisely the same way of writing the name ("Mark J. Nelson"), but only 2 of those are me. If that data were cleaner it'd really increase the usefulness of the profiles. May take quite a bit of magic, though.

mhluongo 14y ago
It does take some magic. Luckily there's been quite a bit of work done of the topic (eg, Lise Getoor & co http://scholr.ly/person/5486566/lise-getoor)- http://scholr.ly/person/5486566/lise-getoor)- we should show some serious improvements soon. In the meantime, I'll manually merge those profiles you've pointed out :) Thanks for taking the time!
- mhluongo 14y ago
  Proper link: http://scholr.ly/person/5486566/lise-getoor http://scholr.ly/person/5486566/lise-getoor

tinco 14y ago

Hi, I tried a search for "model based testing" and the results seem not to be of very high quality. Are you lacking sources? Also, it's not possible to discern high-quality from low-quality papers in any way. As a graduate student, I weed out what I should read by reading the titles+abstracts in the order of the amount of citations. Citation amount maybe isn't the most optimal indicator of quality but it is a strong one. In order of importance: Scope (does it cover all journals?), Quality (how do I know if something's worth reading), Speed (I noticed your search runs a little on the slow side). Other than that the site looks great, I'd certainly switch if you would get the basics down :)

mhluongo 14y ago
Hm, right now we've got the same coverage as CiteSeerX (where most of our data comes from), and are working on expanding. What sources do you particularly think are lacking? We've talked quite a bit internally about whether to include citations counts. It's a tough call, but until we're happier with our scoring it might make sense to help bolster result reputation. Thanks for the feedback!
- tinco 14y ago
  Hey no problem :) I'm just a student so I'm not researching a particular field at the moment, the Model Based Testing paper I was looking for I had to read for a course. I just wouldn't use your site as long as I feel _any_ journal is missing, since google does not give me that feeling.