In order for VSX to be as up-to-date as possible we have to make assumptions.
We usually import variable stars lists containing thousands of objects. We can't check one by one. We do our best to avoid ending up with duplicate records (records matching stars already in VSX) but sometimes the coordinate resolution of one of the two (or both) records is poor (this is typical with sky survey positions) and we can't use a too large radius to cross-identify objects because we may end up with wrong identifications. What is better then? We prefer to delete the duplicates afterwards.
Tamas Zalezsak has been helping us finding more duplicates so we can then merge them and create single primary records.
Survey results are also prone to several problems. Misclassifications are one of them. The automated analysis can't be compared to a knowledgeable human being looking at a light curve (and adding up spectral information too) and the results might be wrong as much as half of the times. Also the classifications they use might be non-standard ("PULS", "MISC", "APER") so this is another area where people (YOU) can help us improve the information included in VSX.
You can submit a revision for any of these survey records including:
1) an accurate position (use UCAC4).
2) a standard variability type (check the VSX variability types document if you need more information)
3) new elements for the variable determined using all the available data (combining ASAS-3, NSVS, SuperWASP, CRTS, APASS and anything else that is available, of course if you have your own observations you can add them too)
4) an improved range or the addition of a range instead of only an amplitude (some surveys only provide a mean or maximum/minimum magnitude and an amplitude). Use standard magnitudes if possible (e.g. if you combine data from the surveys mentioned above, use the APASS/ASAS-3 V zero point, shift the rest of the datasets to that scale and give a V range).
5) any additional information you may find in the literature, e.g.: a spectral type from some VizieR catalogue.
When you submit a revision, make sure to indicate all that has been changed so every user can keep track of what has been done and when those changes took place.
For instance, a typical revision comment would say: "Type, period, epoch and range from ASAS-3, NSVS and CRTS data. UCAC4 position."
Records that should be revised can be identified by means of their positional resolution.
E.g., positions of ASAS records that have not been revised are determined from the ASAS identifiers and those identifiers are such as "ASAS J120010-2334.6" which means that its position in VSX will be 12 00 10.00 -23 34 36.0.
Also you can identify them by the non-standard variability types given.
Checking the Revision history of the star at the bottom of the VSX star page will tell you if a revision has been made or if the object is now the same as it was when it was first imported from the ASAS catalogue.
We also assume that stars published in journals have been checked by the referees that reviewed those papers. Unfortunately that doesn't mean that the published results are okay and we end up finding several misclassifications or wrong periods that are evident at first sight...
Finding these stars with wrong data is more difficult but if you happen to find a paper and you see classifications or periods that don't look okay, know that we welcome revisions of those objects.
So if you are looking for a data-mining project and want to help us with the neverending task of improving the data in VSX, this is a good project because these stars are not a priority to dedicate them staff hours but at the same time it would be great if someone can improve their information.
If you want to take up this challenge and have any doubts, post them here.