Quality of search - Fine-tuning the search

There are many search engines, some are open source and some are paid. The trend in search technology shows an affinity towards cognitive and artificial intelligence now.

• Solr
• Elasticsearch
• Google Search Appliance (GSA)
• Oracle Endeca
• Microsoft FAST
• Attivio
• Sinequa
• Coveo
• IBM Watson
• Amazon CloudSearch
• SharePoint Search
• HP Autonomy

The common Big data used with searches are:

• Apache Hadoop
• Cloudera
• Hortonworks

Quality of search(QoS)
Let us see how we can make a search effective. QoS is a term used with search implementation fine-tuning. Once a search is implemented, the next steps are to ensure the search performance. There are many ways we can do the fine-tuning.

Query cleaning:
Before search hits the index, ensure the query is cleaned and aligned to your index.

Add analytics:
Ensure you are using analytics hand in hand with search, which will definitely improve the effectiveness of search.

Rules:
Rules like ranking, relevance, etc can be adjusted to fine-tune the search results.

Proper metadata mapping:
Re-verify the metadata mapping for the index to ensure everything is done perfectly.

Dictionaries, Synonyms:
Some times we miss to configure dictionaries which will restrict the user from synonym search. Words with similar meaning do not come in result set at all. Confirm you have configured them well.

Excluded words:
Ensure the excluded word list is configured so that none of the unwanted queries fetch the result.

Restrictions:
There are cases where we have to skip some pages from search results. Ensure these restrictions are working as expected.

Once all the above conditions are met, the next step is to verify the search functionality ourselves to confirm the 'recent searches', 'search suggestion', 'auto-complete'/ 'type-ahead', 'null search results' are functioning as expected.

Finally don't miss checking the reports like Terms report, null search report, search requests, index reports, analytics reports, crawl reports, content request reports.

Usually, search technology is treated as 'once implement than ignore'. My recommendation is to keep improving the configurations by constantly watching the reports often.