Needle in a haystack : entity search on text and graph
summary
Combines structured entity data (types and relationships) from knowledge graphs with unstructured Web text to make web-scale entity search more robust. Describes challenges in answering entity-style queries (lists of objects) and an approach that infuses semantic entity information into document text to improve QA-style answers.
AI-written summary · about
chapters (20)
- 0:00Intro
- 1:39What are entities?
- 2:03What is entity search?
- 3:04Problem 1: Lack of coverage
- 3:37Problem 2: Lack of robustness
- 4:15"Knowledge Graph" based Question Answering
- 10:01Connect Web text to Knowledge graph
- 11:39Query Interpretation
- 18:29Query - Web text match (Severyn et. al. 2015)
- 19:38At this stage
- 22:25Latent Variable Discriminative Model
- 24:03Talk Outline
- 28:08Experimental setup
- 29:26Q: is the confluence of KG and Web useful?
- 30:26Examples
- 31:34Q: Comparison with related work
- 32:34Wins/losses
- 33:47Insights
- 35:13Domain-specific applications
- 36:01References
related talks
39mHerding Entities- Information Search and Synthesis in the Context of Transaction Data
22mThe Capabilities of Semantic Search
30mSearch on Steroids — A Story of Semantics
37mBuilding a Search Engine to Improve User Engagement
13mA Beginner’s Guide to Semantic Search
31mAdvanced search for your e-commerce website
41mGetting Started with Query Understanding #bbuzz
37mMachine Learning models utilization