The sentiment analysis block has a second input which allows to exchange the internal sentiment dictionary with a custom one. A sentiment dictionary is used to assign sentiment polarities to terms. In addition, it can be used to assign special meaning to certain terms. A custom dictionary must have the exact same structure as the internal/demo dictionary, i.e. 5 fields in exactly the same order. The internal dictionary is based on the English language and can be found in the demo data section. It is a table in which each row represents a term. The fields have the following meaning:
The actual word. must be spelled exactly as it will appear in the text. If it is a noun, a separate record for the plural should be used as well as all other inflections.
Represents how positive or negative the meaning of a word is. If it is negative, the term describes a bad sentiment. Positive a good sentiment. A polarity of 0 is simply neutral. The terms in the dictionary are either -1, 0 and +1, however it is possible to go higher or lower than -1/+1 or use decimal places to specify stronger / weaker sentiments.
Negations invert the sentiment of a sentence. Values in this field should only be 1 (for negations) or 0 (for everything else).
Terms in this field amplify the sentiment of the context they are located in. E.g. the word "very" in the context "very good" will amplify the positive sentiment of "good", in the same way if will amplify the negative sentiment of "bad" in the context "very bad".
Suppressions are very similar to amplifications but instead reduce the impact of words. A typical example would be "slightly".