Cost saving alternative to custom developed apps for Enterprise Search
Pros
- Index content from many sources including web pages, databases, and RSS feeds with out of the box connectors and customizable index pipelines. Ability to create custom connectors using a JDK.
- Easily modify boosts by various factors to fine tune the relevance of search results.
- Easily capture signals that can be used for popularity boosting and personalization. Allowed us to replace costly and hard to maintain custom applications with out of the box Fusion configurations.
- Easy pre-processing of content at index time and post-processing of content at query time.
- Simplifies Solr configuration management by providing easy access to Solr configuration files and automatic Solr restarts after changes.
Cons
- In the past, documentation did not keep up well with the product but I think that has changed and Lucidworks makes up for that with extensive customer support and professional services when needed.
- Pre-5.0 versions limit machine learning to Spark only models. However, 5.0 and above allows you to embed your own models in other machine learning languages including TensorFlow.
- The upgrade process was cumbersome and time consuming but the scripts they provide have improved with each version. With Fusion 5.0 cloud deployments upgrades should be virtually transparrent.
Return on Investment
- Allowed us to move software developers from enterprise search to other customer focused applications due to built in signals processing and related popularity boosting out of the box.
- Enabled business experts to work directly with Fusion to influence search results and not be so dependent on developers.
