Search is a very complicated technology that can consume a large amount of computing resources. Here we will compare search inside a native application and search on the web.
Search speed on the web, if done correctly, is generally very fast. This is because a) there are a number of very good open-source search engines, b) servers have a lot of RAM, fast I/O and CPU speed. On the other hand, native applications don’t have many open-source search engines, nor do they have the power to search really fast.
The difference probably is even larger when searching through Japanese text. This is because when indexing Japanese, we would generally use
Ngram inflates RAM usage, and is not very realistic to use it on mobile devices.
Server side search is very feature rich
If you use a search engine like Solr, then you can use faceted search. Faceted search tells you the distribution of the results, and you can make it easier for the user to drill down further to specifically find what they are interested in. It would be interesting to see how we can employ faceted search in conferences.
It is not possible to provide the same functionality on a native application, unless that application connects to a server.
Search is vastly better on a client-server architecture with the search engine residing on the server. A native application that searches locally will never have the same level of sophistication and speed will be slower. A better solution would be for the native application to connect to the server whenever the user wants to perform a search.