希望访问中文页面? 请点此(简体中文版)  

Posted on: March 12, 2015

in Blog

​5 Helpful Tips on Running Basic Searches in Relativity

Have you ever had problems with long-running queries in Relativity? Here are some best practices to help make your searches run faster.

5-helpful-tips-for-running-basic-searche

1. If your search only includes a specific range of documents, or documents in a specific tag, always make sure you enter those conditions first.

This will allow the system to cull down to the specified group before running any keywords or metadata conditions. I have also found that by adding all of the “OR” conditions before adding the “AND” conditions, seems to reduce the amount of time it takes to return my results.

2. Use a HASHJOIN query hint if you are running searches on a very large dataset (1 million documents or higher) and the condition to include family.

Enter the hint HASHJOIN:TRUE in the Query Hint box. Do not use this query hint on all searches including family as it can have adverse effects as described below.

When Running Basic Searches in Relativity Use a HASHJOIN Query Hint for Large Datasets

As the hash table size grows, it requires more CPU and memory, so this function should only be utilized in a subset of queries where you are including family. Hash joins deliver the most performance improvement for large datasets when two or more tables are joined together, and one or more of them have no index or are not sorted.

3. Use the lists feature within Relativity when you want to reference another search that has more than one condition.

You do not want to do this on a search that will change as documents are coded or new data is loaded, unless you are certain the list has been updated to the most recent set of documents. However, if you are using a search that is pulling in static information (such as a list of productions) this could be very helpful. It is much faster to add a list condition as opposed to setting the conditions for pulling in all or specific productions.

Below is an example of adding three static conditions that could easily be turned into a list.

Use the lists feature within Relativity when you want to reference another search that has more than one condition.

Make sure your list names are very specific for easy identification. Also, if you are using lists and updating them intermittently, tack on the date to the end of the name.

4. For quality control, always consider adding the fields contained in the conditions to your view.

We all make mistakes and this is a good way to QC your search. If it is not considered an invalid search query, you will get results. When you accidentally join them with an “OR” condition rather than an “AND” condition, you could be looking at irrelevant data. When you add the fields to your view, you are more likely to notice issues, such as information that you intended to exclude appearing in the search results.

5. Use trial-and-error to quality control check searches that contain multiple choice/object fields.

Always QC your results, especially if you are getting specific choices back that you attempted to exclude. Try changing the operator in the conditions from “These conditions” to “Not these conditions”, and then do the same on the popup window and select the opposite to what you had previously. This will fix your result set. If it doesn’t, confirm that you are not pulling in family on the search – this would also cause false/positives to appear in the results.

Below are examples of when you may be in the situation that merits a change in the operator.

From your search conditions view:

Tips for Running Basic Searches in Relativity: Example #1 when you may be in a situation that merits a change in the operator.

Once you click on the browse button from the above condition:

Tips for Running Basic Searches in Relativity: Example #2 when you may be in a situation that merits a change in the operator.

Keeping these best practices in mind while using the saved searches feature in Relativity will help you attain high-quality search results at a faster pace.

Discover More:

Discover More Categories

D4 Weekly eDiscovery Outlook

Power your eDiscovery intellect with our weekly newsletter.