Menu
▾
▴
Re: [Obi-devel] New transformation terms from MO
|
From: William B. <Wil...@Dr...> - 2007-01-27 02:30:33
|
Hi Chris, Thanks for the much more ontologically precise explanation of how these issues are being handled in PATO. PATO quite clearly states on its Wiki page (which I assume you or Giorgio probably wrote) it is "An Ontology of Phenotypic Qualities." It can be difficult to truly grasp how the "monadic" vs. "relational" dialectic applies to some of these measures biomedical scientists would typically contrast as "absolute" vs. "relative". Typically, if a value has units, it is thought of as absolute, where as a unitless number is considered relative. This doesn't hold for everything (such as fundamental constants, some of which are unitless), but it does for most measures. Really that distinction just confuses the matter, I think - I'm sorry I pitched the first email that way. I can remember entering into a debate - possible on the FuGO/OBI list - arguing pH must be "relative" since it's inherently a ratio. Much thought has since gone into this issue amongst PATO developers and BFO users, which now makes the argument you present clearly the way one must approach "values" if you are ultimately expecting to process those facts in an ontologically formal way. Though distance is an absolute measure as most scientists would think of it, unless that number is being dealt with as a number qua number with no relation to the real entities it derived from, as you say, the act of measuring requires reference to more than one entity and therefore is not monadic. I, too, wasn't quite certain how to type correlation coefficients. If they can be said to inhere in a Data_set, one might consider them monadic, but I'm not certain how a Data_set maps into BFO. Correlation does have "relation" in its name for a reason, which seems to imply they must be "relational" qualities. I realize these issues can begin to feel like an "angels on the head of a pin" argument, but I believe the Your contribution on this issue and that of others who will be at the meeting, such as Alan Ruttenberg, will be extremely helpful. I realize these issues can begin to feel like a "number of dancing angels on the head of a pin" argument for some, but I believe you and others can help us better understand how "values" can best be handled by users of OBI. I also believe you, Alan, and others with more BFO experience can provide useful insight on issues related to how OBI currently deals with such entities as "Document", "Software", "Data_set", "Design_specification" - and how these continuants relate to a variety of occurents in OBI. Cheers, Bill On Jan 26, 2007, at 8:12 PM, Chris Mungall wrote: > > On Jan 26, 2007, at 1:21 PM, William Bug wrote: > >> Most definitely, yes. >> >> These are types of quantitative_value which is already in OBI >> (though BFO debates may lead to changing where it ultimately >> resides in the OBI graph). > > Both BFO and PATO agree that there are qualities and that what are > often called "values" are sub-types of more general qualities - eg > square is_a shape. > >> Having said this, you will note the current PATO ontology contains >> a myriad of value types - quantitative, qualitative, relative, >> absolute > > PATO does distinguish between "monadic" and "relational" qualities: > eg the shape of a table is a quality of that table and only that > table; this is a monadic quality. The distance between two objects > is relational - it does not make sense to say the distance is a > quality of object1 without referencing object2 > > these distinctions objective reality and shouldn't be confused with > relative vs absolute which are often artefacts of the measurement. > A long time ago PATO had qualities such as relative_height, > absolute_height, but these refer to the same quality. Now what > confuses most people is the children of these types, such as tall/ > short are themselves relative to some other quality > >> - and subtypes for specific types of values such as speed, color >> intensity, concentration, etc. These are clustered into a "GO >> SLIM" - the OBO format way of creating accessible subsets within >> an ontology. >> >> It may be we'll ultimately want to see OBI values reside in PATO - >> at least, that would be the emerging OBO Foundry way. :-) > > I'm not quite sure what an OBI_value is > > the examples below are interesting. > > whilst the distance metrics can be considered types of qualities > that can inhere in two independent continuants, the others do not > inhere directly in physical objects but rather in variables/data, > themselves dependent continuants. > > let me think about this some more, I'll have an answer by the meeting! > >> There do also appear to be "unit" types in PATO, so this is >> another issue we'd need to sort through re: the developing Units >> Ontology. > > yep - technically a distinct but linked ontology, UO, both > developed under the aegis of PATO > >> Chris Mungall will be present at the OBI meeting, so he can >> definitely help us all to better understand how PATOs values are >> intended to be used by the community. >> >> Cheers, >> Bill >> >> >> >> On Jan 26, 2007, at 3:22 PM, Tina Hernandez-Boussard wrote: >> >>> Hi Ryan and Philippe, >>> >>> Here are some additional data transformation terms from MO. I >>> think that these terms should not go under data transformation, >>> but would be better placed under mathematical terms, as they do >>> not transform the data. >>> >>> Can you please let me know if you are in agreement so that I can >>> get the final list of data transformation term out before Monday. >>> >>> Thanks, >>> >>> Tina >>> >>> >>> NAME:cosine_distanceDEF:The cosine distance of two vectors is the >>> cosine of the angle between them. This measures the difference in >>> direction between two vectors, irrespective of their lengths.MO >>> NAME:Euclidean_distanceDEF:The straight line distance between two >>> points. In n dimensions, the Euclidean distance between two >>> points p and q is square root of (sum (pi-qi)2) where pi (or qi) >>> is the i-th coordinate of p (or q).MO >>> NAME:manhattan_distanceDEF:The "Manhattan distance" is the >>> shortest path between two points in a block format, e.g. the >>> length of the path along Manthattan city streets.MO >>> NAME:pearson_correlation_coefficientDEF:The Pearson's correlation >>> coefficient between two variables. Its values can range between >>> -1.00 to +1.00. The closer the absolute value of the Pearson >>> correlation coefficient is to 0, the smaller the linear >>> relationship between the two variables. A Pearson correlation >>> coefficient with absolute value 1 indicates perfect linear >>> relationship.MO >>> NAME:Spearmans_rank_correlationDEF:Computed as the ordinary >>> Pearson correlation coefficient between two groups of rankings.MO >>> NAME:tau_rank_correlationDEF:a nonparametric measure of the >>> agreement between two rankingsMO >>> NAME:french_railway_distanceDEF:The "French railway distance" is >>> based on the fact that (at least in the past) most of the >>> railways in France headed straight to Paris. Thus, the French >>> railway distance between two points is the usual distance if the >>> straight line through them passes to a designated ?Paris? point, >>> or is the sum of their distances to the ?Paris? point otherwise.MO >>> NAME:jackknife_Pearson_correlationDEF:The jackknife Pearson >>> correlation is the lowest Pearson correlation between two data >>> series where one pair of values in the data series are omitted.MO >>> NAME:uncentered_Pearson_correlationDEF:The uncentered Pearson >>> correlation is defined as the Pearson correlation for two data >>> series where the mean of each data series is assumed to be zero.MO >>> NAME:Pearson_correlationDEF:The Pearson correlation is defined as >>> the covariance of two data series divided by the product of their >>> standard deviations.MO >>> >>> >>> >>> Cheers, >>> >>> Tina B >>> >>> >>> >>> >>> -------------------------------------------------------------------- >>> ----- >>> Take Surveys. Earn Cash. Influence the Future of IT >>> Join SourceForge.net's Techsay panel and you'll get the chance to >>> share your >>> opinions on IT & business topics through brief surveys - and earn >>> cash >>> http://www.techsay.com/default.php? >>> page=join.php&p=sourceforge&CID=DEVDEV______________________________ >>> _________________ >>> Obi-devel mailing list >>> Obi...@li... >>> https://lists.sourceforge.net/lists/listinfo/obi-devel >> >> Bill Bug >> Senior Research Analyst/Ontological Engineer >> >> Laboratory for Bioimaging & Anatomical Informatics >> www.neuroterrain.org >> Department of Neurobiology & Anatomy >> Drexel University College of Medicine >> 2900 Queen Lane >> Philadelphia, PA 19129 >> 215 991 8430 (ph) >> 610 457 0443 (mobile) >> 215 843 9367 (fax) >> >> >> Please Note: I now have a new email - Wil...@Dr... >> >> >> >> > Bill Bug Senior Research Analyst/Ontological Engineer Laboratory for Bioimaging & Anatomical Informatics www.neuroterrain.org Department of Neurobiology & Anatomy Drexel University College of Medicine 2900 Queen Lane Philadelphia, PA 19129 215 991 8430 (ph) 610 457 0443 (mobile) 215 843 9367 (fax) Please Note: I now have a new email - Wil...@Dr... |
