New Search Algorithm for Naver’s Instant Search Results

Naver has enhanced its instant search algorithm with a new function that anticipates users’ search intent.The new technology presents users with pre- set customized search results that provide answers to users’ queries.

According to the search giant, this update is the next step towards their ultimate goal, and one shared by other search platforms, of being able to answer search queries in a natural and conversational way, like a human being.

The new search algorithm is a hybrid blend of instant search, knowledge graph and additional special search features such as currency, weather and calculator features commonly found on other platforms.

For this new feature, Naver focused on further developing its algorithm for NLU (Natural Language Understanding) and combining it with its newly established database called NCR (Naver Content Repository) Mashup.

With the NLU, Naver is trying to understand natural language/sentences as opposed to refined keywords and then matching it up with the related data pulled from the NCR Mashup.

The ‘normal’, single-dimensional search results are currently pulled from several independent and isolated databases. The new NCR Mashup tries to understand the connection between certain search terms and search results from those independent databases and creates a separate database for multi-dimensional search results.

This new algorithm can be best understood by referencing examples. For instance, just by typing the word “highway” into the search field, Naver will still suggest to you the usual instant search results such as, “highway map”, “highway terminal prices”, or, “highway buses”.

In addition to regular instant search suggestions, the new NCR Mashup tries to go one step further by anticipating more complex search intentions and suggests to answer questions such as, “tell me more about the current traffic situation on the highways” (see NCR Mashup box in screenshot below).

13

Similar suggestions are given by just typing in other words such as baseball” which trigger the following NCR Mashup suggestions: “show me the upcoming baseball games”, “show me the current baseball league standings”, and “I want to see the recent baseball highlights”.

Another example of using the NCR Mashup search tool is upon searching for the word, “interesting”, the results read as follows: “recently popular movies”, “recently popular TV shows”, and “recently popular web cartoons”.

Movie- related queries are another example of how this new database tries to understand complex search intensions. If a user watched a good movie and wants to know about other movies showing the same cast, it would require multiple search queries, such as “[actor A] movies”, “[actor B movies]”, etc.

However, when typing in a search query such as “In which other movies does the cast of [movie A] appear?”, Naver will provide a custom designed overview of all actors in that movie and their filmography at one glance (see screenshot below).

25

All of this information was available before the update, but required the user to look up each fact individually or to navigate through the results in order to acquire the desired answer.

With the new NCR Mashup, Korean’s biggest search engine aims to cut down user effort by displaying the desired information right away.

Naver will first implement this new algorithm for search queries in regard to celebrities, movies, and news broadcasting related information, before slowly rolling it out into other areas of the platform.

The ultimate goal is to understand and reply to questions in a way someone would ask and answer in a verbal conversation, with natural sentences rather than keyword- refined search queries.

The Korean search engine summarizes its long-term goals as follows: “From text to speech, from keywords to natural language and from one-dimensional to multi-dimensional search results – this is our dream.”

Hopefully Naver will release some user data or insights to see how many are actually using this new features and whether it really does help the search giant brining it closer to its ultimate goal declared above.

What’s your take on Naver’s new instant search algorithm? We’d love to hear your views and comments below!