Trigger tokens

One simple rule based systems works as follows. Firstly, we require a lexicon which gives the orthography for each word. With this, we can easily determine cheek whether two words in the lexicon have the same orthography, and from this compile a list of tokens which have two or more word forms. Next we write by hand a set of rules which fire (i.e. the activate) on the presence of the trigger token, for example ... 

At run-time, we then move through the sentence left-to-right examining each token in turn. When we find one that is ambiguous (that is, it appears in our list), we look for trigger tokens at other positions in the sentence which will form a collocation with the current token. 

There are fundamental difficulties with regard to this and similar hand written rule approaches. Firstly, it can be extremely laborious to write all these rules by hand. Secondly, we face situations where we have an ambiguous token but no trigger tokens that are present in our collocation list. Both these problems amount to the fact that the mapping from the feature space to the label is only partially defined. Finally, we face situations where say 2 rules match for each word - which do we then pick For these reasons, we now consider machine learning and statistical techniques. 

Those interested in the analysis and modeling of workflow processes have often turned to a process-modeling technique known as the Petri net. This is a popular choice since much work exists that formalizes and enhances the technique introduced by Carl Petri in the 1960s [PN1]. A Petri net is a directed graph that describes the relationship between transitions and the conditions that trigger those transitions to take place. The net executes as transitions consume enabling tokens and produce new tokens that in turn enable other transitions. 

The basic idea of decision lists is to determine the strength of the link between a token and its trigger, and this is done by measuring the frequency of occurrence. In the following table (taken from [507]), we see the raw counts which have been found for the token bass ... 

The double column periphery verifies the trigger by comparing the time stamp with a counter running behind the bunch crossing counter by the trigger delay. In case of agreement the column is set into readout mode and the data acquisition is stopped, otherwise the data are discarded. When the readout token arrives at the double column periphery the validated data are sent to the chip periphery and the double column is reset. The ROCs are read out serially via a 40 MHz analog link. A picture of a BPIX readout chip is shown in Fig. 7.4. 

Personalization With all of this talk about relevance, why not create a subject line that is personalized and speaks directly to the lead If you are using marketing automation, you can create a personalized token, and you can trigger emails based off of behaviors. So if a person makes a purchcise, send her an email saying, Thanks for buying X, we know you will love X ... 

---