The main types of search queries that online store users use, is your website ready for them? (Part 3)
If the correct processing of the requests described in the two previous parts can be considered a prerequisite for site search to be considered “good”, then having coped with the requests given in this part, it will rightfully deserve the title of “excellent”.
In order to provide the user with correct information in response to requests from the first and second part of the article, it is necessary to rely, first of all, on information about a specific product. In such a situation, the main role is played by the name, characteristics, purpose and other information. At the same time, a situation often arises when it is necessary to analyze the request context, user experience, language and other features in order to provide relevant information.
Part 1
Part 2
Using slang, abbreviations and symbols
')
Speaking of slang, we do not offer to sharpen the search under the vocabulary of access culture. We are talking about words that have long been firmly rooted in certain groups and relate to a specific type of product. The closest example is the front derailleur of a bicycle's transmission, which is usually called simply the "front perekidka", respectively, there is also a "perekidka back". If bicycles and spare parts for them are just a small part of the products you offer, then the need to process such a request is small, but if your site is a specialized store, you are guaranteed to receive a large volume of such requests.
In order to be able to effectively handle such requests, you will have to spend a lot of time collecting information about slang that exists among the users of your products. An excellent source will be specialized forums, communities and other thematic resources.
Abbreviations perform a similar role to slang. They often change meaning depending on the context and require special care in processing. Naturally, we are talking about common abbreviations that are understandable to most people interested in this field. A vivid example is the abbreviated name of the English club Manchester United, which is often spelled as “MU”, moreover, such a reduction is often found in colloquial speech. At the same time, specialized shops of football paraphernalia can give an adequate response to such a request only if the abbreviation is found in the name, that is, almost never.
By the way, returning to the slang, requests such as "Gunners" or "Red Devils" also do not give any results.
Often the problem of context arises with symbols. The dash dash may denote a range of values (15-17 "), a negative value (-10 degrees), or an attempt to exclude a word from the search results, by analogy with most search engines.
Omissions
Even if a person “communicates” with a machine, he still remains a person who has errors and omissions. This is not a typo, the incorrect processing of which is genuinely surprising, but that the user does not indicate a part of the information, taking it for granted. You can efficiently process such requests taking into account information about a specific user that you have. Starting from the source of the transition, ending with the history of visits to your site
.
For example, if a user navigates to an advertising site that says that a certain brand’s products are being sold on your site, it is logical to assume that when typing the word “jacket” in the search box, he is looking for the jacket of the particular manufacturer specified in the advertisement. You should not overdo it, limiting the search results will reduce to zero the possibility of spontaneous buying a jacket from another manufacturer. Therefore, it is necessary either to display all the results, place the jackets of the brand specified in the advertisement first, or use a filter by brand, which, if turned off, will allow viewing the entire range.
Colloquial
Problems with this type of query are also due to the difference between how the program perceives the request and how the person formulates it. The use of colloquial speech in queries is a rather rare phenomenon, but it is starting to gain momentum. Their processing requires the synthesis of several search technologies in order not only to define keywords, despite what word form they were used in, but also the general context, as well as the priority of each of the words.
The query "red women's shoes 38-39", in principle, contains all the necessary information in order to correctly display the results that will be useful to the user. But at the same time, the search algorithm should contain, at a minimum, a connection between the word “shoes” and, suppose, two-digit numbers (if you use the European size system). That is, contextual restrictions must be set that exclude incorrect interpretation of the information specified in the request.
Naturally, today there is no objective need for extremely correct processing of this type of requests. Most users have sufficient experience in using various search engines, which allows them to quite effectively adjust their queries to achieve an optimal result. But at the same time, the development of personalization, an individual approach and the desire for the intuitiveness of the interface of online stores greatly “relax” the visitor, who not only can and will use the types of requests described above, but may well consider their incorrect processing as a reason for search for another store.
In order to provide the user with correct information in response to requests from the first and second part of the article, it is necessary to rely, first of all, on information about a specific product. In such a situation, the main role is played by the name, characteristics, purpose and other information. At the same time, a situation often arises when it is necessary to analyze the request context, user experience, language and other features in order to provide relevant information.
Part 1
Part 2
Using slang, abbreviations and symbols
')
Speaking of slang, we do not offer to sharpen the search under the vocabulary of access culture. We are talking about words that have long been firmly rooted in certain groups and relate to a specific type of product. The closest example is the front derailleur of a bicycle's transmission, which is usually called simply the "front perekidka", respectively, there is also a "perekidka back". If bicycles and spare parts for them are just a small part of the products you offer, then the need to process such a request is small, but if your site is a specialized store, you are guaranteed to receive a large volume of such requests.
In order to be able to effectively handle such requests, you will have to spend a lot of time collecting information about slang that exists among the users of your products. An excellent source will be specialized forums, communities and other thematic resources.
Abbreviations perform a similar role to slang. They often change meaning depending on the context and require special care in processing. Naturally, we are talking about common abbreviations that are understandable to most people interested in this field. A vivid example is the abbreviated name of the English club Manchester United, which is often spelled as “MU”, moreover, such a reduction is often found in colloquial speech. At the same time, specialized shops of football paraphernalia can give an adequate response to such a request only if the abbreviation is found in the name, that is, almost never.
By the way, returning to the slang, requests such as "Gunners" or "Red Devils" also do not give any results.
Often the problem of context arises with symbols. The dash dash may denote a range of values (15-17 "), a negative value (-10 degrees), or an attempt to exclude a word from the search results, by analogy with most search engines.
Omissions
Even if a person “communicates” with a machine, he still remains a person who has errors and omissions. This is not a typo, the incorrect processing of which is genuinely surprising, but that the user does not indicate a part of the information, taking it for granted. You can efficiently process such requests taking into account information about a specific user that you have. Starting from the source of the transition, ending with the history of visits to your site
.
For example, if a user navigates to an advertising site that says that a certain brand’s products are being sold on your site, it is logical to assume that when typing the word “jacket” in the search box, he is looking for the jacket of the particular manufacturer specified in the advertisement. You should not overdo it, limiting the search results will reduce to zero the possibility of spontaneous buying a jacket from another manufacturer. Therefore, it is necessary either to display all the results, place the jackets of the brand specified in the advertisement first, or use a filter by brand, which, if turned off, will allow viewing the entire range.
Colloquial
Problems with this type of query are also due to the difference between how the program perceives the request and how the person formulates it. The use of colloquial speech in queries is a rather rare phenomenon, but it is starting to gain momentum. Their processing requires the synthesis of several search technologies in order not only to define keywords, despite what word form they were used in, but also the general context, as well as the priority of each of the words.
The query "red women's shoes 38-39", in principle, contains all the necessary information in order to correctly display the results that will be useful to the user. But at the same time, the search algorithm should contain, at a minimum, a connection between the word “shoes” and, suppose, two-digit numbers (if you use the European size system). That is, contextual restrictions must be set that exclude incorrect interpretation of the information specified in the request.
Naturally, today there is no objective need for extremely correct processing of this type of requests. Most users have sufficient experience in using various search engines, which allows them to quite effectively adjust their queries to achieve an optimal result. But at the same time, the development of personalization, an individual approach and the desire for the intuitiveness of the interface of online stores greatly “relax” the visitor, who not only can and will use the types of requests described above, but may well consider their incorrect processing as a reason for search for another store.
Source: https://habr.com/ru/post/238457/
All Articles