Friday, 4 September 2015

Remembering John Greaves

John Greaves is one of only a tiny number of UK amateur astronomers who had a peer-reviewed article published in one of high profile astronomical journals. (Mon. Not. R. Astron. Soc. 355, 585-590 (2004)).

John hasn't been, as far as I know, ever a member of any astronomical society but at one time he played a significant role in both astronomical data mining and in variable star astronomy.

John has never been afraid to “shoot from the hip” and this example of a John diatribe is typical of his work.  

<Quote starts> 

Given the moribund state of the AAVSO Data Section and little history of data mining work or publication by the current section leader whilst Martin has :-

i) a history of engaging in data mining exercises, both with online data sources and using robotic telescope systems (ie similar to AAVSOnet)
ii) run webpages and websites outlining astronomical projects for years, including projects that deal with variable stars
iii) self confessed mentored beginners in his variable star datamining projects
iv) has had published in peer reviewed journals, such as OEJV and the AAVSO's own JAAVSO, works based on his data analysis of online epoch photometry data sources
v) been the second highest contributing poster to the current AAVSO Data Section mailing list after the current Data Section leader himself, with Martin giving advice comments and suggestions, whilst Michael just mostly gives pep talks with little advice or guidance, as Michael's specialist astronomical interests are more geared towards doing his own observing and analysing his own data.  The lag in time the moderation process takes between Michael receiving an email and posting it to the group does tend to stifle any potential dialogue
vi) and given the lack of direction and guidance to the section, whereas Martin has created and ran several similar groups in the past it seems quite logical, if Martin is an AAVSO member, for his expertese to be utilised by this stalled Section.  Not only stalled, but examination of its online archives shows that requests for assistance and advice made to it have gone unheeded, or worse non-datamining projects suggested (once or twice to the potential benefit of AAVSO but not to furthering an individual's datamining skills or best practice).

It must be stated though that I am not an AAVSO member.  I have some small experience in data analysis and the publishing of the results therefrom in refereed journals, and I was somewhat instrumental in explaining and advising which of the first datasets imported into AAVSO VSX after the GCVS and NSV were likely to be useful, and explaining their format and how to convert and/or what their fields meant to Chris Watson so he could import them, as well as one or two other little niceties re AAVSO VSX.  So I am not disconnected from the topic nor unfamiliar with the matter.

Possession of the above demonstrable credentials: maintaining astronomical mailing lists; running webpages with astronomical projects based on datamining; experience and use of _AAVSO_ VSX data and data submission; publishing of peer reviewed papers on datamining including within the AAVSO's own journal; and mentoring on data mining, is unparallelled by any other AAVSO member currently.

If the running and maintaing of astronomical mailing lists is removed from that list, then one other AAVSO member would fulfil the criteria.  If further removing the running of webpages on datamining from the list, then that adds one other extra AAVSO member.  Even with only the remaining credentials, especially the core two of having peer reviewed publications upon datamining (extra especially within JAAVSO itself) and of advising and directing datamining projects, then _only_ these three people within AAVSO membership have the recorded credentials, and in fact the majority of _AAVSO VSX moderators_ cannot fulfil more than one of the criteria, least of all the peer reviewed publishing of papers using datamining, and that includes the ones that are AAVSO salaried staff.

Thus given the above credentials there is but one logical consequence.

Meanwhile, the current AAVSO Data Section head moderates the group at an absentee level, with postings lagging up to a week before being passed on, thus stymieing any potential dialogue.  That's the maintaining a mailing list criterion.  There is nothing wrong with moderating a list, as long as the list is maintained in a timely manner, and that responsibility maintained.  Although Michael Koppelman maintains astronomical webpages, Slacker Astronomy is primarily an act of journalism.  Although Michael has published it is primarily a case of publishing his own observations, not datamining work, and more recently as a professional collarborator with other professionals, not so much as a pro-am, least of all in an amateur datamining based collaboration.

His leads on the archived Data Section list have been few, mostly asking others for help and advice and suggestions of what to do, which is strange for a leading and directing role.  His comments are mostly to support comments made by Arne.  The science advisor Doug Welch tends to just utter "well done" supporting homilies, his only major suggestion being to recommend the MACHO database to people, yet without highlighting to beginners how to handle this problematic resource (throwing out a couple of general references is not the same as giving a walkthrough when dealing with beginners, this is supposed to be a support group.  I've some familiarity in this area, as I've mentored data mining for variable stars in MACHO data that has led to refereed publications by others).

AAVSO officials have even presented and strongly recommended exercises that are not datamining exercises, thus potentially misleading the novice as to what datamining actually is.  Arne has suggested people wade through the GCVS looking for objects noted as "different" and informing him of them.  This he could readily do himself with the most basic filtering of B/GCVS at VizieR.  Neither was any mentoring provided to assist beginners in knowing what is "different", no list of preferred pathologies or phenomena itemised, not even a note of preferred category of variability.  The Section leader merely backed up Arne's call with no further input.  Neither is such a task a datamining task, nor does it further the capabilities of the novice.

Michael Simonsen, another AAVSO staffer, merely suggested that Data Section members import CRTS epoch photometry into the AAVSO International Database.  This too is not datamining, teaches no skills, highlights no analyses, and benefits no one wishing to learn dataminig.  It merely bloats AAVSO International Database with yet another publicly available dataset, letting it appear that AAVSO is the repository of all known observational data by mirrorring said.  Except for it doesn't, because Michael S. seemed completely unaware when he made the suggestion that the CRTS epoch photometry is in fact _not_ publicly available, albeit there being long term plans to make it so, at which time it will likely be downloadable en masse as a dataset (given the other database server that group has produced as precedent) and best imported into AID via scripts in one go, now the AID is itself an SQL database.  And of course, he was only referring to the transients, the cataclysmic variables, a notoriously problematic subgroup of variable stars in terms of datamining, their being aperiodic erratics with no true outburst patterns most of the time, not even at quasicyclic levels.  So, no datamining there, just a plumping up of AID's contents.  The Z Cam campaign has some merit (and sounds somewhat familiar), but has at best been a 'datamining-lite' exercise, despite the fact that many erroneous UGZ classifications can be traced back to an early 1960s paper which made suggestions of that subclass for some stars (eg AB Dra) based on a then paucity of data, and it can thus be shown that there was no reason for future publications to take these classifications up as demonstrated fact (nor did the original publication particularly affirm that.  Basic literature work via online resources, an inherent and essential aspect of variable star datamining, readily reveals this.  The paper is not in English, this might be part of the problem.

When it comes to APASS data release 0, Arne mentioned that AAVSO Data Section analysed the data for him.  In fact, Patrick Wils analysed the data, and replied on the AAVSO Data Section list where the request was made, ie where the mailing thread lived.  No one else from AAVSO did any analysis of the data at all, and Patrick Wils did his analysis within the context of being Patrick Wils.  However, this is irrelevant, for that exercise was not the datamining analysis of epoch photometry.  Granted datamining is not restricted to the analysis of epoch photometry, but publication in this field shows that for variable stars it is the predominant result generating activity in terms of datamining.  The analysis of variable stars is most frequently that of the data available upon them, which in available archive terms consists predominantly of epoch photometry.  No doubt some excuse will be used that APASS intends to form a calibrating reservoir for variable star epoch photometry.  This is still not datamining.  This is still not teaching people how to datamine.

Contrast that to the credentials listed above and there is a logical consequence of whom in AAVSO is currently most suited to steer a datamining section via the actual record of publishing achievement and relevant practices.


PS Incidentally, who is actually giving and running the actual core datamining workshops advertised in the programme for the Argentine meeting, as it does not say?  Is it the AAVSO Data Mining Section, or is it an individual?  Possibly the individual is a member of the AAVSO Data Mining Section so it will be claimed to be being given by the Data Mining section, in the same way Patrick Wils' personal assessment of APASS data release 0 was claimed to be an analysis conducted by the AAVSO Data Mining Section, thus incidentally not attributing proper credit for effort expended and work done.  And no doubt Michael Koppelman and/or Doug Welch will stand up and include this in any statement of AAVSO Data Section's achievements since the last meeting.  These are the current leadership's credentials, compared to the abovemost enumerated credentials.

< Quote ends>


No comments:

Post a Comment