What are the ways to play search operations, do you know?

Introduction to is the continuous optimization and iteration of technical students. It is easy to overlook the role and value of operating students who can directly reach the business side. So today, I will share with you what are the gameplays of operating students in search? Alibaba Cloud Open Search is a one-stop visual search development platform. What optimization actions can product/operation students participate in on the console?

Reading time: 5 minutes

Suitable readers: search operation positions, product positions, and people who are interested in search technology;

search is a must-have function for every product, and it is also the most important item for business growth. products with richer information will pay more attention to search capabilities, and in the search scenario, users "actively" generate content , which requires search Technology can accurately analyze recalls and hit search intent to achieve the purpose of business conversion.

Most of the products on the market are searched by technical students who are constantly optimizing and iterating. It is easy to overlook the role and value of operating students who can directly reach the business side. So today, I will share with you what are the gameplays of operating students in search? Alibaba Cloud Open Search is a one-stop visual search development platform. What optimization actions can product/operation students participate in on the console?

What kind of operations can be done on search?

1. Cooperate with product technology, be responsible for search indicators, and continue to follow up the iteration of search capabilities;
2. Output search evaluation reports by analyzing business indicators, formulate optimized solutions, and improve product capabilities and user experience;
3. Make breakthrough explorations of user experience from the product and operation level, combine search guidance function with operation plan, and enhance business conversion;
4. In accordance with laws and regulations, control the security of content and deal with the bad information content in the platform in a timely manner;

Difficulties in operating classmates

1. Lack of search-related technical knowledge reserves, technical concepts and logic are not clear, which is not conducive to the development of operations and the advancement of projects;

• How do rewriting, word segmentation, recall, and sorting work in search, and what optimizations can operations participate in?
• What are the ways to play personalized search?
• How to optimize search to improve user business realization?

1. demand resolution cycle is long, it is difficult to respond promptly and quickly, and business development is slow;
2. Enterprises do not have complete data management capabilities, and operations cannot view search business index data in real time to make corresponding operational analysis and operational decisions;

• Core search operation data: search, traffic, behavior, transaction, user analysis, query analysis, etc.;
• Personalized search guidance: data analysis such as drop-down prompts, hot words, shading, etc.;

Interpretation of search core logic

Search business process:

1. Interpret the information entered by the user

Query semantic understanding function description : "translate" the query entered by the user into a meaning that the system can understand, realize effective communication between humans and computers, and analyze the user's search intent, recall the most relevant content/commodity, and resolve User search request.

Each function in query semantics understanding will directly affect the analysis of user search intent and the recall effect, thereby directly affecting business indicators such as click-through rate, bounce rate, and conversion rate. At the same time, it is also necessary to continuously optimize and explore operations, products, and technologies based on their own product conditions.

Example: Searching for "aj1 North Carolina New Sneakers", the computer did the following query analysis and processing

2. Filter user intent relevant content

After the user’s query is interpreted, some standardized words will be obtained, which correspond to related content, and the screening of content involves two concepts: recall rate and accuracy rate.

• Accuracy refers to the proportion of relevant content in the searched content;
• Recall rate refers to the proportion of content that is actually searched out.

What is a recall? performs word segmentation through the keywords that the user queries, and quickly locates the segmented phrase to the document by searching the inverted linked list. This process is called recall;

When the ratio of these two indicators is closer to 1, the effect is better. However, in some cases, accuracy and recall are a set of contradictory indicators. For example, only one search result is searched and it is the real intention of the user. That is accurate The rate reached 100%, but the recall rate was very low. These two concepts are key indicators in search optimization and involve more advanced search mechanisms.
Note: Not all results containing user query keywords should be recalled.

3. Sort the search results

After the query analysis is recalled, the content/commodities that are most in line with the user's intentions are sorted reasonably to increase the click-through rate and prevent users from jumping out. Next, we will introduce the sorting rules of the search.

• Coarse sorting : Perform the first round of selection of search results. Because you need to traverse all the documents, the coarse sorting should be as simple as possible (select the most important items of the document, such as news, you can choose text and timeliness ), the documents are scored according to the expression and sorted according to the score result.
• fine ranking: of the first round of coarse ranking results and performs the second round of more detailed score calculation according to the fine ranking, and performs the final sorting according to the score, and returns it to the user.
• sorting expression: used to control the sorting of search result documents. It supports basic operations (arithmetic operations, relational operations, logical operations, bit operations, conditional operations), mathematical functions, and sorting features. The sorting expression can be used for in-depth tuning of the sorting effect.

4. Personalized search peripheral functions: (focus on operations!!)

4.1 Hot search shading

hot search shading is a basic function necessary for a complete search engine. It usually occupies an important position at the entrance of the search box and provides indispensable business value . Located at the most upstream of the entire workflow of the search engine, it provides a pavement for search optimization, which can greatly reduce the difficulty of optimization in query understanding, sorting, and operation intervention, and it can have a larger space for combined operation strategies.

From the user's point of view, the hot search shading can generally meet the following needs:

1. I want to stroll around, I don’t know what to search, can you recommend some high-quality query words for me?
2. I want to know what everyone has searched, you can’t go wrong with the crowd
3. It’s best to recommend queries based on my interests, and have diversity. I want to see what I’m interested in, but I also want to explore some content outside of my interests.

From the operator's point of view, hot searches and shading can provide such value:

1. I want to know which queries are searched the most. Popular queries are the indicator of user interest. By analyzing popular queries, we can grasp the trend of users’ interests and provide decision-making basis for formulating operational strategies.
2. I want to recommend some high-quality queries to users. When the user has input, the drop-down prompt guides the user's intentions, but when there is no input, how can I recommend high-quality queries?
3. If you recommend popular queries to users, you can’t always give the hottest queries. Diversity needs to be considered. On the one hand, user experience needs to be taken into consideration. On the other hand, it is necessary to expose some of the most popular queries
4. By analyzing the user’s behavior and combining the user’s interests to recommend queries, it not only takes into account the user experience, but also improves the business goals in a targeted manner.

✅ operations can focus on business indicators:

hot search:

• Hot Search PV : The number of hot search requests (and return success) that day;
• hot search UV : the number of users who requested hot search on that day;
• hot search UV-CTR : user clicks on the hot search results;
• Hot Search PV-CTR : Hot Search Hits
• guided search PV-CTR : Measure the recall and sorting effects of hot search guided searches;
• guide search GMV : measure the effect of hot search guide purchase;
• guide search favorite/comment/like conversion rate : measure the hot search guide favorite/comment/like effect;

Shading:

• shading PV : the number of times that shading was requested (and returned successfully) on that day;
• Shading UV : the number of users who have requested shading on the day;
• Shading UV-CTR : the user clicks on the shading result;
• shading PV-CTR : the click situation of shading;
• guide search PV-CTR : measure the recall and sorting effect of shading guided search;
• guide search GMV : measure the effect of shading to guide the purchase;
• guide search collection/comment/like conversion rate : measure the effect of shading guide collection/comment/like;

4.2 drop-down prompt

The drop-down prompt is the basic function of the search service. In the process of the user inputting the query, the candidate query is intelligently recommended to improve the input efficiency of the user and help the user find the content they want as soon as possible. You can search for the drop-down prompt candidate query by Chinese prefix, Pinyin Quanpin, Pinyin first letter simplified Pinyin, Chinese character plus Pinyin, prefix after word segmentation, Chinese homophones, etc.

✅ Operation can focus on drop-down reminder business indicators:

• pull-down prompt PV : the number of times that the pull-down prompt was requested (and returned successfully) on that day;
• pull-down reminder UV : the number of users who requested a pull-down reminder that day;
• drop-down reminder PV-CTR : the click situation of drop-down reminder, to measure the effect of drop-down reminder recall and sorting;
• Pull down pull reminder UV-CTR : the user clicks on the result of pull down reminder, to measure the effect of pull down reminder recall and sorting;
• guide search GMV : drop-down prompt to guide the search transaction amount;
• guide search for favorites/comments/likes conversion rate : measure the effect of drop-down prompts to guide favorites/comments/likes;

Open search platform introduction

OpenSearch (OpenSearch) Alibaba Cloud is a one-stop intelligent search business development platform built by a self-developed large-scale distributed search engine. High-quality search services can be obtained with one-click access without development. Built-in Alibaba technology has accumulated for many years The core search engine, industry-leading search capabilities and algorithm capabilities , and fully open to support internal calls to customers’ own algorithm models to meet the business needs of various industries and scenarios, and achieve mutual achievement and common growth with customers;

Operational highlights and advantages:

1. Visual console, clear modules, friendly to novices, easy to operate, no need to wait for the development cycle, convenient for non-technical operating product personnel to optimize effects at any time according to business conditions;
2. Industry-leading technology creates unique industry search templates, one-click configuration, built-in industry search capabilities, and high-quality search capabilities for industry attributes without development and training;
3. Support developers' algorithm models to be returned online in real time, and model development can be superimposed on existing platform capabilities according to their own business conditions;
4. Supports A/B Test, so that a certain percentage of traffic can be allocated a priori before the full amount of service is used, so as to avoid the negative impact on online business caused by blind usage.
5. Support cloud monitoring and alarm, through cloud monitoring to monitor the application's storage capacity, computing resources, query QPS and other indicators. Help monitor the usage of the application, and support the setting of alarm rules for monitoring items to keep abreast of business trends.
6. Support customized search services without the need to form a technical team to solve business difficulties, top Alibaba technology and algorithm engineering teammates will help you solve problems;

Operation/product operable menus and functions

Application management related

1. View : application list, application details, application specifications, application cloud monitoring;

2. Operation : second-level expansion and contraction , changing specifications, easy to deal with big promotions and other activities;

Search Algorithm Center

1. View: All menus can be viewed

2. Recall configuration :

a. Word segmentation management:

i. Test word segmentation effect

ii. Add, delete, modify the word segmentation entry of custom word segmentation

b. Query analysis:

i. Search test

ii. Configure query analysis rules : rewrite strategy and function selection

Configuration rewriting strategy: The terms that can be controlled to participate in the recall are included in the query results in an AND or OR relationship. rewriting of 160f531096722f directly affects the results of the recall and can be adjusted according to business conditions.

Example: If the query is: "Nike sports shoes", the term after the word segmentation is: "Nike/sports/shoes"

• • When the connector is AND: the result of Query rewriting is (default:'Nike' AND default:'sports' AND default:'shoes')
  • When the connector is OR: the result of Query rewriting is (default:'Nike' OR default:'sports' OR default:'shoes')

Query and analysis function selection: Select the industry template by default to select all.

c. Dictionary management:

Add, delete, and modify the intervention entries of each functional dictionary: spelling error correction, stop words, synonyms, entity recognition, word weight, category prediction intervention

3. Sorting configuration:

a. Sorting strategy management: configure sorting expressions according to business needs to optimize sorting effects

b. Search test: you can view the score results of each function

4. Search guide function:

a. Drop-down prompt:

According to the data characteristics of different industries, the drop-down prompt function provides corresponding optimized templates. Currently, templates for general, e-commerce and content industries are supported;

i. View the report related to the drop-down prompt, effect of 160f531096734f

ii. Select configure the black and white list

iii. Personalized configuration : high-frequency search terms, user search frequency is used as the basis for ranking candidate words, and the words with higher search frequency of users meeting the recommendation conditions are given priority as the drop-down prompt candidate query. Historical search terms, give priority to the query that the user has searched before. Intelligent sorting, intelligently sort candidate queries based on user behavior information such as clicks and purchases.

b. Hot search and shading:

i. View reports hot search shading 160f531096739e, preview of effects

ii. Select configure the black and white list

iii. Effect optimization -Behavioral data: The hot search/shading results based on the statistical search log can meet the needs of the cold start phase. After this function starts to work, it is recommended to associate the hot search/shading user click events, the system Internally, the effect is further optimized by collecting behavioral data. The collected behavioral data has the following benefits:

• Statistics can be obtained for various indicators, such as the pv, uv, and non-result rate of the guided search, which are used to measure the effect of this function and provide a basis for subsequent improvements.
• It can analyze the interest trend of the user group and provide a basis for formulating operation strategies.
• Intelligent methods can be used to recommend queries, and data can be annotated through user click behaviors. The model may be trained according to different optimization goals (optimized by click-through rate by default), and queries can be recommended through the model, with strong generalization ability .
• You can make personalized recommendations for popular search terms. Once you know which queries the user has clicked, you can make targeted recommendations based on the user's preferences.

​

Statistical report

all reports can be viewed: business operation report, drop-down prompt report, hot search shading report, A/B test report;

All of the above menu and function introduction, operation steps and precautions can be found in open search product documentation .

---

If you need product guidance, you can fill out the questionnaire to get expert guidance \>> https://survey.aliyun.com/apps/zhiliao/lKD\_J8cRj

Copyright Statement: content of this article is contributed spontaneously by Alibaba Cloud real-name registered users. The copyright belongs to the original author. The Alibaba Cloud Developer Community does not own its copyright and does not assume corresponding legal responsibilities. For specific rules, please refer to the "Alibaba Cloud Developer Community User Service Agreement" and the "Alibaba Cloud Developer Community Intellectual Property Protection Guidelines". If you find suspected plagiarism in this community, fill in the infringement complaint form to report it. Once verified, the community will immediately delete the suspected infringing content.