The following dictionary attributes either affect the way Zing indexes tables or ranks search results.

Index and search dictionary attributes

Zing uses dictionary attributes set at the table or field level to configure the following settings:
  • indexing behavior
  • ranking of search results
  • the BM25 relevancy ranking algorithm
Table 1. Zing index and search dictionary attributes
NameData TypeApplies ToApplies to V3 or V4DescriptionLabelNeeds re-indexing?Root Table Only
attachment_index boolean Table Both If true, attachments on the table are indexed for search purposes. Attachment Index TRUE FALSE
no_text_index boolean Both Both This attribute on a field prevents the field from being included in a text index. No Text Index TRUE FALSE
text_index _attachment_body _weight float Table V4 Specifies how much consideration (weight, a float value) to give a search term found in the body of an attachment. For more information on how weight value affect search result relevancy, see Zing computes document scores using three components. Text Index Attachment Body Weight FALSE TRUE
text_index _attachment_displayed boolean Table Both Displays attachments in text search results. Text Index Attachment Displayed FALSE FALSE
text_index_attachment _title_weight float Table V4 Specifies how much consideration (weight, a float value) to give a search term found in the title of an attachment. For more information on how weight value affect search result relevancy, see Zing computes document scores using three components. Text Index Attachment Title Weight FALSE TRUE
text_index _auto_stop boolean Table Both Automatically identifies stop words and doesn't index them. Based on the threshold specified in text_index_auto_stop_threshold. Text Index Auto Stop TRUE TRUE
text_index_auto _stop_threshold int Table Both Threshold (integer, number of occurrences) for a word to become a stop word automatically. See text_index_auto_stop. Text Index Auto Stop Threshold TRUE TRUE
text_index_default _partial_matching_rule String Table Both String for partial matching rule configuration. Input should be based on Change the query mode of an indexed table. Text Index Default Partial Matching Rule FALSE TRUE
text_index _default_query_mode String Table Both String for query mode configuration. Input should be based on Change the query mode of an indexed table Text Index Default Query Mode FALSE TRUE
text_index _disable_synonym boolean Table Both TRUE disables searching on synonyms as well as the search term. Text Index Disable Synonym FALSE TRUE
text_index _enable_idf boolean Table V3 TRUE enables TF-IDFscoring. Text Index Enable IDF FALSE TRUE
text_index _filter_junk boolean Table Both TRUE by default, FALSE disables the junk filter for a table. By default, Zing doesn't index or search for 2-digit numbers and single character words (unless they are Chinese or Japanese characters). Regenerate the index after disabling the junk filter. This attribute results in a larger table index. For optimal performance, don't apply it unless it's required. Text Index Filter Junk TRUE TRUE
text_index _filter_query An encodedQuery on the table Table Both Use this attribute when you want to index the rows matching specific filter conditions. The attribute value must be an encodedQuery on the table. Index Filter TRUE FALSE
text_index_filterable boolean Column Both If a field with this attribute set to true, is changed, the row is evaluated for re-indexing (even if the field itself isn't indexable). Index Filterable TRUE FALSE
text_index_include _stop_word boolean Table Both if you need to index stop words in this table, set this attribute to true. Text Index Include Stop Word TRUE TRUE
text_index_scorer_bm25_b float Table V4 This attribute is the b parameter used in the BM25 algorithm. See text_index_scorer_bm25_b. bm25_b FALSE TRUE
text_index_scorer_bm25_k1 float Table V4 This attribute is the K1 parameter used in the BM25 algorithm. See text_index_scorer_bm25_k1. bm25_k1 FALSE TRUE
text_index_tags String Table Both Should be set to specific String: empty/everyone_only/all_shared, depending on type of tags that wants to be indexed. Text Index Tags TRUE FALSE
text_index_tokenizer _language String Table Both Put "ja" (without double quote) when there is need to support tokenization in Japanese. Text Index Tokenizer Language TRUE TRUE
text_index_translations boolean Table Both If present or true, forces indexes to be recalculated when translated strings are added. Requires sys-admin role to modify. Automatically set for indexed fields that are translated, and to fields that have a translation and are being indexed. The glide.i18n.force_index system property, which defaults to true, overrides this attribute. Text Index Translations TRUE FALSE
text_search_and_or_limit int Table Both Add table level custom attribute text_search_and_or_limit that overrides the global property. Text Search And Or Limit FALSE FALSE
ts_language String Column Both Added to allow for text indexing of only specific languages. Text Search Language TRUE FALSE
ts_multi_value_field _splitter String Column Both If one field is a multi-value field, add the splitter in the field in this attribute to spread out index positions of the values. For example, to separate multi:value, put : for this attribute. Text Search Multi-value Field Splitter TRUE TRUE
ts_weight int Column Both Controls the relative importance of a match in the field for text search. For more information on how weight value affect search result relevancy, see Zing computes document scores using three components. To learn how to set text search weight for a field, see Set the relative weight of a field. Text Search Weight FALSE FALSE

no_text_index

For Customer Service and Support guidelines on setting the no_text_index option for different field types, see KB0859922 on the Now Support site.

text_index_scorer_bm25_b

This dictionary attribute is part of the BM25 algorithm. The attribute impacts the importance of the length of a document in search results. If this attribute is 0, the length has no bearing. In this case, a document of two words would be valued equally as a document with 1000 words. Very short documents are often weighted lower than longer documents. The default is 0.75.

text_index_scorer_bm25_k1

This dictionary attribute is part of the BM25 algorithm. The attribute helps determine term frequency saturation characteristics, that is, having the search term appear in a document increases the relevancy of the document, but the increase is less impactful as the numbers climb. For example, a search term occurring 10 times in a document makes it more relevant than a document that has only one occurrence. However, a document that has 100 occurrences of the search term may not be ten times more impactful than a document that has 10 occurrences. The relevance approaches a maximum effect asymptote. The higher the attribute value, the lower the maximum effect of multiple occurrences. Typical values are between 1.25 - 2.0.