Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.ai    |    Awaiting the gospel from Sarah Connor    |    1,954 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 1,038 of 1,954    |
|    Dmitry A. Kazakov to Market Theory    |
|    Re: Algorithms/code for discovering patt    |
|    12 May 06 09:40:03    |
      XPost: comp.programming       From: mailbox@dmitry-kazakov.de              On Fri, 12 May 2006 01:09:12 GMT, Market Theory wrote:              > pete wrote:       >> Market Theory wrote:       >>>       >>> A variant of the General Poster's problem.       >>>       >>> Discover patterns of consecutive symbols in a stream. The patterns are       >>> not known in advance, so this is a statistical anomaly detection       >>> problem. The stream consists of mostly random symbols over a known       >>> finite alphabet. The expected frequencies/probabilities of the random       >>> symbols may be known in advance, or inferred from the stream. The       >>> minimum and maximum length of patterns of interest are specified as       >>> parameters.       >>>       >>> eg       >>>       >>> wnckbvmnunixwjgemblsmznhauiwriewpklgnkwnbwnbwknqjowjunixjowrgnblsmknashwiw       >>>       >>> has the pattern "unix"       >>       >> ... and also "blsm"       >       > Well spotted! but what was your algorithm?              I bet it was a dictionary created from the sample. Then the frequency of       each token in the dictionary was compared with its expected frequency,       evaluated from the frequencies of individual symbols of each token.              A more realistic problem stating would also take into account correlations       between consequent symbols. So "blsm" would get a much lesser expected       frequency, because of four consonants in a row. (Assuming a natural       language source, rather than a cat walking on the keyboard.) Which would       position "blsm" far before "unix"! (:-))              --       Regards,       Dmitry A. Kazakov       http://www.dmitry-kazakov.de              [ comp.ai is moderated. To submit, just post and be patient, or if ]       [ that fails mail your article to |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca