Case Study #3 Wesco

  1. Solved data duplicates issue which client was facing while using Solr
  2. Motivated client to purchase LW Fusion and migrated search solution from Solr to LW Fusion
  3. Help client to setup LW Fusion 4.x in Linux environment
  4. Created collections and modified Solr config files as per need
  5. Developed Query pipelines – (query design, relevancy logic & signals implementation)
  6. Developed Index pipelines
  7. Reduced data indexing time
  8. Created custom dashboard to show zero result queries, top searched terms, clicked item not in first page, etc.
  9. Knowledge sharing with client on delivered tasks

Lucidwork Fusion - Wesco

Lucidwork Fusion - Wesco

  1. Only developer with a LW architect
  2. End-to-end development of ecomm data search usecase
    • Index design – SKU, Customer Part Number and Typeahead
    • Index pipelines
    • Query pipelines
  3. Search Flow
    • Multi tiered search flow with CPN call, spellcheck flow and zero results handling
    • Custom recall based on user segments
    • Fields based relevance, custom boosting and signal boosting
    • JSON facets and dynamic facet logic
  4. Typeahead
    • Based on CPN, Manufacturer and Categories fields
  5. Rules
    • Business Rules
      • Set Facets based on term and category path
      • Redirect
    • Rewrite Rules
      • Synonyms
      • Spellcheck
      • Phrase
  6. Jobs
    1. Click signal aggregation
    2. Typeahead indexing
    3. Synonyms job to move rewrite rules to synonyms.txt
  7. Documentation

Lucidwork Fusion - Wesco

  1. Provided general suggestion on client’s questions related to best practices
  2. Developed POC on below:
    • A/B test using dummy signals generated from signals
    • NER implementation to identify if searched term contains Manufacturer name, and if found then boost results for specific Manufacturer
    • Item recommendations based on BPR algorithm in LW Fusion available as OOTB feature
    • Item recommendations based on ALS algorithm in Custom PySpark code developed using Jupyter
  3. Bug fixes on search issues
  4. Created custom job to fetch details of clicked item not in first page and results with no click
  5. Discussed the possible scenarios to implement search use cases using ML with LW Fusion
  6. Knowledge sharing with client on delivered tasks and documented the same

Thank You

For more information about us, Contact us at:

Nextbrick Logo

Let's break ice

Email Us

For Search, Content Management & Data Engineering Services

Get in touch with us