...
| Key | Usage | Default |
|---|---|---|
| max_contributing_items | Number of contributing items (definitions) that are considered for the Smart Filter training. Only the latest entries are considered. | 20 |
| min_feature_count | Number of times a feature (extracted token) needs to be present in the training text before it is included in the Smart Filter. The default value (0) causes this to be calculated automatically. By setting it to a non-zero value a minimal number of occurrence is enforcedSee Improving Relevance in the reference for background information on this setting. | 0 |
| Phrase Detection | ||
| phrase_detection | Whether phrase detection is enabled. Set to "false" to disable. Phrase detection tries to extract multi-word tokens. As an example then instead of extracting two terms "income" and "statement", the Smart Filter might contain of one term "income statement". | true |
| phrase_length | How many terms are included in phrases. | 2 |
| max_phrases_per_language | The maximum number of phrases that are extracted for any language. | 5 |
| min_phrase_count | Number of times a phrase has to be present for it to be included in a Smart Filter. | 2 |
...