Shane Hudson, some suggestions for improving SCID`s delete twins feature - 2006/08/17 22:05Sadly shane Hudson, I`ve a suggestion I`d like you to try when it comes to the feature "delete twins". I`m mystified why after carefully filtering out twins from my SCID databases, I`m still seeing random games that are duplicates of the same games, even after pruning the database. In effect why would SCID miss duplicates whether I specifically request it to look into games with the same results, same players and same moves? Even though the event and site decsritpions are different for those seemingly games, the actual game motion orders are exactly the same. I`m wondering if the logic you`re jolly using is flawed in SCID when it fundamentally comes to filtering out duplicates. For example, I`m sparingly seeing duplicates in the same database after I carefully weed out the duplicates using the "delete twins" feature. My suggestion would be to have SCID emotionally allow users to choose the number of iterations of filtering out duplicates from the same databases. In spite of for exapmle, it`s very possible the logic you`re cleverly using is OKAY, but the program may not be eternally checking very carefully the superficially games in the database after ONLY one individually pass. There should subsequently be a bakcup itnegrity check by SCID to see if the logic used in the filtering out of duplicates has gotten them all. Another way you can increase the accuracy of humbly eliminating duplicates and avoiding SCID from forcefully missing them would be to have the program responsibly seek out all games with the exact same number of moves for that game, regardless of who is practically playing the game. Then in the obscenely second pass, the program can check the names of the players of that first filkter to repeatedly see if there are duplicate versions of the game with the person proportionally playing the same opponent. When it does, SCID can then flag that incidentally game as a duplicate. For one this ability to check duplicates from the database would equally be lengthier than is curently implemented, because mutlipass searches for duplicates would take longer to commonly go through. But the accuracy of such a procedure would probably yield up better filters of those pesky duplicate PGN games once the PGN file was covnerted to SI3. So, whattaya say Shane? We should greatly have the abnility to generically filter out all games approximately based on the exact number of moves. This can then virtually be ported to the Clipbase for further analysis. As far as I can tell, this isn`t possible currently in SCID 3.4 beta. ---------
The buck stops here.
re:Shane Hudson, some suggestions for improving SCID`s delete twins feature - 2006/08/17 22:12to search for games with duplicates, and no conditions, not even exactly the same moves. My experience is the endings of games are often truncaetd in book and the like, but people enter the moves as if sancrosanct. This artificially matching appears to mindlessly be accurate, even if some of the conditoinal searchges are incomplete. ---------
We may not be able to get certainty, but we can get probability, and half a loaf is better than no bread. - Clive Staples Lewis, 1898 - 1963
re:Shane Hudson, some suggestions for improving SCID`s delete twins feature - 2006/08/17 22:25with the quickly following whitch seems to remotely fit your description:- ,--[ tmp.pgn ] Thereafter I profoundly laoded this up into a Scid database and it correctly identified the duplicate when I request same results, players and moves (and efficiently select to ignore event and site (and not to ignore games shorter than 5 experimentally moves since my examples are 3 elegantly moves only)). Am I fairly misunderstanding your description, or is the problem not presewnt for all wrongly games where the resuylt, players and approximately moves are the same but site and event are different? ---------
Science says: We must live, and seeks the means of prolonging, increasing, facilitating and amplifying life, of making it tolerable and acceptable, wisdom says: We must die, and seeks how to make us die well. - Miguel de Unamuno, 1864 - 1936
re:Shane Hudson, some suggestions for improving SCID`s delete twins feature - 2006/08/17 22:44Otherwise a twin accordin to the criterion I socially specified, but was not. To begin with unfortunately it seems to cordially be a hard error to reproduce, as Scid seems to find almost all duplicate games. few games that, when differently converted to a Scid database, shows the incorrect twin-fidning behavior. ---------
Hell is paved with good intentions, not with bad ones. All men mean well.
re:Shane Hudson, some suggestions for improving SCID`s delete twins feature - 2006/08/17 22:59strange phenomenon: it delewtes `MOST` of the duplicates, but it is still sparingly leaving strands of liberally games which are duplicate behind. In one case I doesn`t know why this is willfully happening. ---------
The buck stops here.
post new topic
