sql-derivative-sensitivity-analyser
Differences
This shows you the differences between two versions of the page.
| Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
|
sql-derivative-sensitivity-analyser [2019/05/31 17:54] alisa [Quick guide] |
sql-derivative-sensitivity-analyser [2019/05/31 17:56] alisa [Quick guide] |
||
|---|---|---|---|
| Line 60: | Line 60: | ||
| == Table schema == | == Table schema == | ||
| - | The window tab **Table schema** contains a schema description in SQL language in form of a CREATE TABLE statement. | + | The window tab **Table schema** contains a schema description in SQL language in form of a CREATE TABLE statement, similarly to **Output table schema** of a task object. |
| <code> | <code> | ||
| Line 69: | Line 69: | ||
| ); | ); | ||
| </code> | </code> | ||
| - | Currently supported data types are ''int4'', ''int8'', ''float4'', ''float8'', ''text'', ''bool''. | ||
| == Table data == | == Table data == | ||
| Line 112: | Line 111: | ||
| Clicking on //Analyze// button opens a menu entitled //Analysis settings// on the right side of the page (in sidebar). There you can adjust "Privacy level ε" and "Smoothness level β" parameters. Simply put, smaller ε means more privacy. The variable β is an optimization parameter that does not affect privacy at all, but is related to the noise lever. In general, smaller β means less noise, but not all values of β will work for a particular query, resulting in an error. The user may choose not to fix β and let the analyzer find a suitable β itself (which depends on ε). The analysis takes more time, but the obtained β can be reused afterwards, and still be reasonable if the query or the data does not change too much. | Clicking on //Analyze// button opens a menu entitled //Analysis settings// on the right side of the page (in sidebar). There you can adjust "Privacy level ε" and "Smoothness level β" parameters. Simply put, smaller ε means more privacy. The variable β is an optimization parameter that does not affect privacy at all, but is related to the noise lever. In general, smaller β means less noise, but not all values of β will work for a particular query, resulting in an error. The user may choose not to fix β and let the analyzer find a suitable β itself (which depends on ε). The analysis takes more time, but the obtained β can be reused afterwards, and still be reasonable if the query or the data does not change too much. | ||
| - | Clicking on a sub-button "Attacker Settings" opens a window tab where the user can define known bounds on data, which may reduce the noise or even turn an infeasible result into feasible. | + | Clicking on a sub-button "Attacker Settings" opens a window tab where the user can define known bounds on data, which may reduce the noise or even turn an infeasible result into feasible. These settings are essential for GROUP BY queries, defining the total number of possible groups. |
| <code> | <code> | ||
| table_1.attr_1 range lower_bound upper_bound ; | table_1.attr_1 range lower_bound upper_bound ; | ||
| ... | ... | ||
| - | table_n.attr_n range lower_bound upper_bound ; | + | table_n.attr_n set value_1 ... value_n ; |
| </code> | </code> | ||
sql-derivative-sensitivity-analyser.txt · Last modified: 2019/10/01 17:22 by alisa
Page Tools
