Complete exclusion of categories when searching

Short description

Allowing users to select and exclude certain categories when searching can encourage honest product descriptions without the need for moderator reviews. Trust between search engines, resource owners, and users has declined in the online community, and improving quality requires new mechanisms and monetization models. Large e-commerce platforms like Amazon and AliExpress have search algorithms that are based on keyword and product category searches, and live sellers have an advantage over search engines as they exclude unnecessary categories and do not assign unnecessary keywords to products. The method of category exclusion can also be used for navigating social networks. The Toronto School of Communication offers new communication technologies and protocols as an alternative to server-based platforms and AI.

Complete exclusion of categories when searching

Adding functionality that allows users to select certain categories and completely exclude others when searching can encourage sellers and site owners to honestly describe the categories and properties of products or resources without the need for moderator reviews or user complaints.

The level of mutual trust can vary from community to community. Trust is influenced not only by developed social connections, but also by useful communication practices and protocols. In the online community, the level of trust between search engines, resource owners and users has only been declining since the turn of the century. The owners of the resource try to screw up the rank in the search engine output, and the owners of the search engine fight this by changing the algorithms of the search engine in a hidden way. This article describes the possibility of creating interaction protocols between the search engine, resource owners and users, in which the rank of a search query increases in the presence of correct metadata provided by the resource owner to the search system, and decreases in the presence of incorrect or excessive metadata solely due to the structure of the search query, and not due to verification by moderators or hidden search engine verification mechanisms.

ChatGPT’s articles discussing the threat of Google search refer to the list of companies Killed by Google, but the place for Google’s grave has long been determined by itself, its commercial model is incompatible with quality search and quality presentation of information. In order to be profitable, the search result must allow for inconspicuous advertising.If Google gives you the perfect answer to your query, you won’t click on any ad link“. The use of ChatGPT and other artificial intelligence systems is affected by the monetization model. Improving quality requires not only new mechanisms, but also a monetization model that stimulates improvement. Such a model is available in large e-commerce platforms: AliExpress, Amazon, Ozona, Yandex Marketa, Wildberries and the like.

When searching the Internet, users use keywords, and sometimes exclude some of them. In large e-commerce platforms, searches are based on keywords and product categories. Usually, the keyword exception does not apply, except for Yandex Market. Search engines try to simplify the interface and ensure the naturalness of search requests from the user, including the ability to ask questions in natural language. However, such requests and the simulation of a “reasonable human response” by AI may contradict the real practice of communication between a buyer and a live seller. A live seller can be a technical language translator at the buyer’s request. However, if the buyer and seller have a good understanding of the topic, their language of communication will be “less natural”. For example, at the request of the buyer to find a bolt with an M12 thread and a hole for a cotter pin, the seller will immediately exclude unnecessary categories from their search and will not discuss bolts without a hole or cotter pins like Google or ChatGPT. The seller also has an advantage over the search engine in that his store catalog for internal use is not cluttered with unnecessary keywords and product categories. Similarly, “honest” are electronic catalogs on the sites of individual stores and navigation of an individual store on a platform, for example, on AliExpress, where there is no need to assign unnecessary keywords and categories to a product in order to promote it in the search against other products of the same store.

The implementation of the search algorithm by a live seller in an “honest” Internet catalog can be organized in the following way. The first step in the search is to search by keywords. The second step is not going through useful results or clarifying the selection of categories based on the search results, but complete exclusion of some categories to which the useless results of the first step of the search belong[1]. For example, in the list of categories in the left column, each category should have two mutually exclusive check boxes – one to select this category, the second to completely exclude all objects belonging to this category from the search results. The ability to select, search, and navigate categories encourages resource owners to provide categories. The ability to exclude unnecessary categories encourages resource owners to honestly assign only necessary categories to avoid falsely excluding unnecessary, albeit close, categories from search results. Categories can also be displayed on a separate picture as intersecting sets, rather than just as a list with a picture of a typical representative. Each domain in the image can be selected or excluded. Implementation of the search algorithm by a live seller in an “honest” Internet catalog

The category exclusion method only works if the resource owner is actually promoting their resource to interested users, for example, selling real products, and not trying to spam certain searches. The method itself works in e-commerce, but requires some additional mechanisms to combat spam in another network, for example, the mechanisms discussed in the B++ Manifesto. The method of excluding categories can also be used when navigating in social networks, where categories can be applied to posts by users or authors.

Examples: Combining keyword searches with category selection doesn’t work well

Let’s try to find it on Yandex Market women’s cashmere cardigan. The result contains non-cardigans and non-cashmere cardigans. Excluding keywords to refine the result is useless. Navigation by category also does not work well: select “Clothing and shoes / Women / Jumpers, sweaters, cardigans”, and then Type = cardigan, Composition = cashmere. After that, there is an extremely small selection of first positions, in which at the time of writing the article is occupied by cardigans with a composition of “acrylic 50%, wool 50%”, but cardigans with 100% cashmere are also available. Cardigan “Viscose 52%, Polyamide 21%, Polyester 27%”, which was ranked first in the keyword search, was rightly excluded from the category search results. Yandex-market either inefficiently processes the “Stock” field, or promotes products that do not belong to the search category, or allows sellers to choose categories themselves and abuse this choice, not taking into account the actual value of the “Stock” field in the product description.

If the user or the sellers do not have a good idea of ​​the category and name of the product he is looking for, the search is even more difficult. Let’s give a simple example in English[2]: search for a service on Alibaba integrated circuit packaging, i.e. housing of silicon crystals in the housings of integrated circuits by the wire bonding method. In such a search, it is necessary to exclude the services of “integrated circuit” sellers, as well as the services of packing goods into boxes and boxes “packaging”, including packaging with wire binding, although a Google image search wire bonding indicates mainly the method of welding contacts when encapsulating microcircuits. With an open service request integrated circuit packaging on Alibaba or search terms, most of the results will be from IC and electronic component wholesalers, some results from PCB manufacturers and assemblers, some from packaging services and from packaging manufacturers. Deadlines cannot be excluded for an effective request on the platform integrated circuit and packagingbecause the exclusion of any of these terms will exclude integrated circuit packaging quite. The solution to the problem is to exclude categories of goods and service providers, that is, categories integrated circuit, packaging and other non-search categories as they are discovered, but not keywords.


This article is written within the framework of the development of the ideas of the Toronto School of Communication and the application of these ideas to solving practical problems of the Internet. The Toronto School of Communication studies existing human communication technologies and offers new communication technologies (protocols). As part of the critique of communication protocols, I analyzed how existing communication protocols lead to the failure of scientific theories. The approach of the Toronto School of Communication assumes that new communication technologies of a large number of people are able to influence people’s thinking and develop “collective minds” taking into account new protocols, such as B++. Peer-to-peer “collective intelligence” protocols should be considered an alternative to server-based platforms and AI.

[1] Refining a query by excluding keywords cannot replace category exclusions, so the same keywords can be present in the description of mutually exclusive categories, and the product description can contain words used in the description of categories that do not belong to this product. Instead of excluding keywords, you need to exclude categories, but the mechanism for excluding categories is not presented in search engines. E-commerce platforms show a list of categories or allow navigation by categories belonging to the products found, which allows you to deselect some of them, but leaves the problem of excessive or careless categorization.

[2] This example does not translate well into Russian due to the narrower term “korpusvania” used in Russian, instead of the broader term “packaging” when translated directly into English.

If you look events in Mykolaiv –

Related posts