With the rapid development of business, more and more customers have higher and higher requirements for the actual transformation and distribution efficiency brought by search. Especially in e-commerce, information and other industries, this goal is extremely important.
Ali Cloud OpenSearch is a search engine solution for the whole industry. However, due to the complexity of search requirements in different fields and business scenarios, the current algorithm function of a single fixed template type cannot effectively solve the problem of personalized search effect optimization among different customers. Based on this situation, Ali Cloud OpenSearch will release a series of functions such as data collection, AB test and algorithm intervention in mid-September, thus opening the era of personalized search.
I believe that we will have such experience, in Taobao search time will always find the front of the goods are often their own heart expected that result, sigh Taobao know my heart. And behind this is actually a large number of accurate depth algorithms work together to produce the effect. In order to build a set of accurate search algorithm, it normally requires at least three technicians proficient in algorithm to spend more than a year, after countless model training and debugging can have certain results. However, ali Cloud OpenSearch has the same powerful algorithm as Taobao search without knowing the complicated and obscure algorithm logic after the data acquisition service and AB test function are launched.
According to the introduction, after users open the data collection service, they can synchronize the relevant user behavior data of their products to the OpenSearch database through the SDK of OpenSearch. At that time, when users use the algorithm function of OpenSearch (such as category prediction), they can choose to use the uploaded behavioral data for model training. In this way, the trained model is closely related to the actual business scenarios and user needs, which is bound to greatly improve the transformation effect of search. However, the model is not a one-off and the rationality of parameter configuration needs debugging and polishing. Therefore, after model training, users can use the newly released AB test function to extract part of the online random flow for experiment, and compare the transformation effect of the online historical model, so as to determine the next optimization decision. For badcase that cannot be solved efficiently by means of parameter adjustment, users can manually intervene badcase through the newly launched algorithm intervention function, so as to quickly stop losses from online problems.
The release of this new generation of functions of Ali Cloud OpenSearch will more deeply meet the needs and visions of search engine distribution data and information efficiency from individual developers, small and medium-sized enterprises to large-scale enterprise businesses, and open the personalized era of cloud search services.
Learn more big data family product details, welcome to click: et.aliyun.com/bigdatarele… Click for big data family product launch: yq.aliyun.com/webinar/pla… Ali cloud launch 】 【 way for the development of new generation of intelligent data: yq.aliyun.com/roundtable/…