Fuzzy Lookup Transformation
Applies To: SQL Server 2016
The Fuzzy Lookup transformation performs data cleaning tasks such as standardizing data, correcting data, and providing missing values.
The Fuzzy Lookup transformation differs from the Lookup transformation in its use of fuzzy matching. The Lookup transformation uses an equi-join to locate matching records in the reference table. It returns records with at least one matching record, and returns records with no matching records. In contrast, the Fuzzy Lookup transformation uses fuzzy matching to return one or more close matches in the reference table.
A Fuzzy Lookup transformation frequently follows a Lookup transformation in a package data flow. First, the Lookup transformation tries to find an exact match. If it fails, the Fuzzy Lookup transformation provides close matches from the reference table.
The transformation needs access to a reference data source that contains the values that are used to clean and extend the input data. The reference data source must be a table in a SQL Server database. The match between the value in an input column and the value in the reference table can be an exact match or a fuzzy match. However, the transformation requires at least one column match to be configured for fuzzy matching. If you want to use only exact matching, use the Lookup transformation instead.
This transformation has one input and one output.
Only input columns with the DT_WSTR and DT_STR data types can be used in fuzzy matching. Exact matching can use any DTS data type except DT_TEXT, DT_NTEXT, and DT_IMAGE. For more information, see Integration Services Data Types. Columns that participate in the join between the input and the reference table must have compatible data types. For example, it is valid to join a column with the DTS DT_WSTR data type to a column with the SQL Server nvarchar data type, but invalid to join a column with the DT_WSTR data type to a column with the int data type.
You can customize this transformation by specifying the maximum amount of memory, the row comparison algorithm, and the caching of indexes and reference tables that the transformation uses.
The amount of memory that the Fuzzy Lookup transformation uses can be configured by setting the MaxMemoryUsage custom property. You can specify the number of megabytes (MB), or use the value 0, which lets the transformation use a dynamic amount of memory based on its needs and the physical memory available. The MaxMemoryUsage custom property can be updated by a property expression when the package is loaded. For more information, see Integration Services (SSIS) Expressions, Use Property Expressions in Packages, and Transformation Custom Properties.
The Fuzzy Lookup transformation includes three features for customizing the lookup it performs: maximum number of matches to return per input row, token delimiters, and similarity thresholds.
The transformation returns zero or more matches up to the number of matches specified. Specifying a maximum number of matches does not guarantee that the transformation returns the maximum number of matches; it only guarantees that the transformation returns at most that number of matches. If you set the maximum number of matches to a value greater than 1, the output of the transformation may include more than one row per lookup and some of the rows may be duplicates.
The transformation provides a default set of delimiters used to tokenize the data, but you can add token delimiters to suit the needs of your data. The Delimiters property contains the default delimiters. Tokenization is important because it defines the units within the data that are compared to each other.
The similarity thresholds can be set at the component and join levels. The join-level similarity threshold is only available when the transformation performs a fuzzy match between columns in the input and the reference table. The similarity range is 0 to 1. The closer to 1 the threshold is, the more similar the rows and columns must be to qualify as duplicates. You specify the similarity threshold by setting the MinSimilarity property at the component and join levels. To satisfy the similarity that is specified at the component level, all rows must have a similarity across all matches that is greater than or equal to the similarity threshold that is specified at the component level. That is, you cannot specify a very close match at the component level unless the matches at the row or join level are equally close.
Each match includes a similarity score and a confidence score. The similarity score is a mathematical measure of the textural similarity between the input record and the record that Fuzzy Lookup transformation returns from the reference table. The confidence score is a measure of how likely it is that a particular value is the best match among the matches found in the reference table. The confidence score assigned to a record depends on the other matching records that are returned. For example, matching St. and Saint returns a low similarity score regardless of other matches. If Saint is the only match returned, the confidence score is high. If both Saint and St. appear in the reference table, the confidence in St. is high and the confidence in Saint is low. However, high similarity may not mean high confidence. For example, if you are looking up the value Chapter 4, the returned results Chapter 1, Chapter 2, and Chapter 3 have a high similarity score but a low confidence score because it is unclear which of the results is the best match.
The similarity score is represented by a decimal value between 0 and 1, where a similarity score of 1 means an exact match between the value in the input column and the value in the reference table. The confidence score, also a decimal value between 0 and 1, indicates the confidence in the match. If no usable match is found, similarity and confidence scores of 0 are assigned to the row, and the output columns copied from the reference table will contain null values.
Sometimes, Fuzzy Lookup may not locate appropriate matches in the reference table. This can occur if the input value that is used in a lookup is a single, short word. For example, helo is not matched with the value hello in a reference table when no other tokens are present in that column or any other column in the row.
The transformation output columns include the input columns that are marked as pass-through columns, the selected columns in the lookup table, and the following additional columns:
_Similarity, a column that describes the similarity between values in the input and reference columns.
_Confidence, a column that describes the quality of the match.
The transformation uses the connection to the SQL Server database to create the temporary tables that the fuzzy matching algorithm uses.
When the package first runs the transformation, the transformation copies the reference table, adds a key with an integer data type to the new table, and builds an index on the key column. Next, the transformation builds an index, called a match index, on the copy of the reference table. The match index stores the results of tokenizing the values in the transformation input columns, and the transformation then uses the tokens in the lookup operation. The match index is a table in a SQL Server database.
When the package runs again, the transformation can either use an existing match index or create a new index. If the reference table is static, the package can avoid the potentially expensive process of rebuilding the index for repeat sessions of data cleaning. If you choose to use an existing index, the index is created the first time that the package runs. If multiple Fuzzy Lookup transformations use the same reference table, they can all use the same index. To reuse the index, the lookup operations must be identical; the lookup must use the same columns. You can name the index and select the connection to the SQL Server database that saves the index.
If the transformation saves the match index, the match index can be maintained automatically. This means that every time a record in the reference table is updated, the match index is also updated. Maintaining the match index can save processing time, because the index does not have to be rebuilt when the package runs. You can specify how the transformation manages the match index.
The following table describes the match index options.
|GenerateAndMaintainNewIndex||Create a new index, save it, and maintain it. The transformation installs triggers on the reference table to keep the reference table and index table synchronized.|
|GenerateAndPersistNewIndex||Create a new index and save it, but do not maintain it.|
|GenerateNewIndex||Create a new index, but do not save it.|
|ReuseExistingIndex||Reuse an existing index.|
The GenerateAndMaintainNewIndex option installs triggers on the reference table to keep the match index table and the reference table synchronized. If you have to remove the installed trigger, you must run the sp_FuzzyLookupTableMaintenanceUnInstall stored procedure, and provide the name specified in the MatchIndexName property as the input parameter value.
You should not delete the maintained match index table before running the sp_FuzzyLookupTableMaintenanceUnInstall stored procedure. If the match index table is deleted, the triggers on the reference table will not execute correctly. All subsequent updates to the reference table will fail until you manually drop the triggers on the reference table.
The SQL TRUNCATE TABLE command does not invoke DELETE triggers. If the TRUNCATE TABLE command is used on the reference table, the reference table and the match index will no longer be synchronized and the Fuzzy Lookup transformation fails. While the triggers that maintain the match index table are installed on the reference table, you should use the SQL DELETE command instead of the TRUNCATE TABLE command.
Because the Maintain stored index option requires CLR integration, this feature works only when you select a reference table on an instance of SQL Server where CLR integration is enabled.
When you configure the Fuzzy Lookup transformation, you can specify the comparison algorithm that the transformation uses to locate matching records in the reference table. If you set the Exhaustive property to True, the transformation compares every row in the input to every row in the reference table. This comparison algorithm may produce more accurate results, but it is likely to make the transformation perform more slowly unless the number of rows is the reference table is small. If the Exhaustive property is set to True, the entire reference table is loaded into memory. To avoid performance issues, it is advisable to set the Exhaustive property to True during package development only.
If the Exhaustive property is set to False, the Fuzzy Lookup transformation returns only matches that have at least one indexed token or substring (the substring is called a q-gram) in common with the input record. To maximize the efficiency of lookups, only a subset of the tokens in each row in the table is indexed in the inverted index structure that the Fuzzy Lookup transformation uses to locate matches. When the input dataset is small, you can set Exhaustive to True to avoid missing matches for which no common tokens exist in the index table.
When you configure the Fuzzy Lookup transformation, you can specify whether the transformation partially caches the index and reference table in memory before the transformation does its work. If you set the WarmCaches property to True, the index and reference table are loaded into memory. When the input has many rows, setting the WarmCaches property to True can improve the performance of the transformation. When the number of input rows is small, setting the WarmCaches property to False can make the reuse of a large index faster.
At run time, the Fuzzy Lookup transformation creates temporary objects, such as tables and indexes, in the SQL Server database that the transformation connects to. The size of these temporary tables and indexes is proportionate to the number of rows and tokens in the reference table and the number of tokens that the Fuzzy Lookup transformation creates; therefore, they could potentially consume a significant amount of disk space. The transformation also queries these temporary tables. You should therefore consider connecting the Fuzzy Lookup transformation to a non-production instance of a SQL Server database, especially if the production server has limited disk space available.
The performance of this transformation may improve if the tables and indexes it uses are located on the local computer. If the reference table that the Fuzzy Lookup transformation uses is on the production server, you should consider copying the table to a non-production server and configuring the Fuzzy Lookup transformation to access the copy. By doing this, you can prevent the lookup queries from consuming resources on the production server. In addition, if the Fuzzy Lookup transformation maintains the match index—that is, if MatchIndexOptionsis set to GenerateAndMaintainNewIndex—the transformation may lock the reference table for the duration of the data cleaning operation and prevent other users and applications from accessing the table.
You can set properties through SSIS Designer or programmatically.
For more information about the properties that you can set in the Fuzzy Lookup Transformation Editor dialog box, click one of the following topics:
For more information about the properties that you can set in the Advanced Editor dialog box or programmatically, click one of the following topics:
For details about how to set properties of a data flow component, see Set the Properties of a Data Flow Component.