Figure 4 Sequence diagram model with context-awareness information and process variables to ubiquitously perform output results.
Estimates (mean and accompanying 95% credibility interval) for the sixteen sets of base frequencies at the ancestral root sequence under the context-dependent model with context-dependent model frequencies.
Figure 9 shows (log) Bayes Factors for our context-dependent model with context-dependent model frequencies and Markov chains of different orders at the ancestral root sequence (see also Table S5 in Additional file 1).
As mentioned above, the context-dependent model with context-dependent model frequencies and a second-order Markov chain at the ancestral root sequence yields a a bidirectional mean log Bayes Factor of 7221.47 log units, compared to the independent GTR model (7218.01 ([7200.34; 7235.68]) log units for the annealing scheme and 7224.94 ([7207.16; 7242.71]) log units for the melting scheme).
We provide additional information on this approach in Additional file 1. Figure 6 (a) shows a comparison of the substitution patterns obtained using our context-dependent model with context-dependent model frequencies, with both a first-order Markov chain and the first-order successive approximation approach to model the ancestral root sequence.
We have calculated (log) Bayes Factors for our context-dependent model with context-dependent model frequencies and Markov chains of different orders at the ancestral root sequence in the same way as reported above, with the general time-reversible model as the reference model.
Estimates (mean and accompanying 95% credibility interval) for the four sets of base frequencies at the ancestral root, for both first-order Markov chain approach and the successive approximation first-order approach using our context-dependent model with context-dependent model frequencies.
However, not much work has been focused on high-level event modeling with context information, such as spatio-temporal information.
A framework for log-linear models with context specific independence structures, i.e. conditional independencies holding only for specific values of the conditioning variables is introduced.
In this context, the work also presented opportunities in which emerging technologies such as RFID, which carry a high potential as tracking data sources, would significantly enhance the modeling efforts by provisioning the models with context-specific empirical inputs as close to real time as possible.
