How to get superior text processing in Python with Pynini
Regular expressions are the standard for string processing, but did you know you can often get better text untangling with Pynini's finite-state transducers?
Latest from O’Reilly
Insights, practical guidance, and announcements from O'Reilly
Growing a design team
5 questions for Alastair Simpson: Customer empathy, culture’s impact on design, and the power of saying “I don’t know.”
Scaling oneself: Delegating as a leader means letting go
It’s important to be willing to delegate tasks to the experts who can best help achieve the company vision.
Design.bio
Bringing biology to the design studio.
Learning How to Delegate as a Leader
This report examines the best ways to delegate work so you not only end up with successfully completed tasks, but also a team of cheerful people—including a cheerful you.
How containers add to a proven cloud-native architecture
Mike McGarr and Andrew Spyker explain the potential containers have to help Netflix create a more productive development experience while simultaneously deepening its control over resource management.
Understanding reactive architecture through the actor model
The O’Reilly Podcast: Hugh McKee on learning how to think asynchronously to create highly concurrent business systems.
A guide to Elasticsearch 5.0 and the ELK/Elastic Stack
Learn how to ship, parse, store, and analyze logs.
Using Linux tracing tools
Sasha Goldshtein reviews the modern tracing landscape—ftrace, perf_events, kprobes, uprobes, eBPF—uncovering what insight into system activity these tools can offer.
Service virtualization, an enterprise solution to legacy systems
How replacing costly mainframes with virtual assets speeds the testing process, reduces time to market, and increases agility.
Lessons learned from building bots
Andy Mauro discusses pitfalls, opportunities, and the future of conversational bot interactions.
Repeatable sampling of data sets in BigQuery for machine learning
An efficient, fast, and repeatable selection method that works on very large data sets.