This tool is only available if you have an active subscription with Professional or Team pricing plan.
Select a pricing plan

Solution for working with semantic kernels

Semantic kernel

This module allows quick and convenient creation of semantic kernels with tree-like structures and automatic clustering for websites or advertising campaigns.
As you can see from the name, this tool allows to prepare semantic kernel of a website or advertising campaign for further operation.

Main features of this tool are keywords clustering, visual node arrangement and building a tree of categories and sections for a website and its articles.
You begin working with this tool by creating a semantic kernel. To do this click the "New project" button on https://www.keys.so/clustering page

When creating a project you should specify a source of initial keywords for the project. Any of your group reports, a list of keywords downloaded from a text file, or reports specified as a list can serve as such source.

After clicking the "New project" button, the system will prepare the environment for operation.

Using keywords unknown to keys.so
If you import keywords from files or a list, the system will download all specified keywords into your kernel, but search volume and clustering will only be available for keywords present in the system. You can process the rest of keywords manually.

You can also download keywords with their own search volume values, the file format for download: Keyword;Search volume;Exact search volume. Example of a file

If a keyword is already in the kernel, its search volume data will be updated.
Module interface
The main window is divided into two areas. To the left is the group tree, by default it includes only the two system items – Non-grouped and Deleted. You can create new nodes, as well as change or move existing ones.
Tree nodes
The group control panel has several buttons:
1
Filtration field
The Filtration field allows to quickly filter displayed nodes. With this tool you can quickly find a folder you need. This filter has the special "+" operator, which is used to search for list items that contain several fragments of text . For example, in a semantic kernel for a cooking website you can specify "dough+dumplings" and only nodes containing the "dough" and "dumplings" words in any order in their names will remain visible. In our case these may be tree nodes and articles like "dough for dumplings" or "making dumplings without dough".
2
Additional filtering button
When you click this button, the modal window will open. In this window, folders can be additionally sorted by the following parameters:

  • Total number of keywords
  • Search volume of the top keyword
  • Total search volume
3
Navigate to folder
When you click this button, the tree view area will scroll to bring a selected folder into view.
4
New node button
When you click it, a child node will be created for your currently selected node. This button is active on any non-system node, i.e. you won't be able to create embedded nodes in "Non-grouped" and "Deleted" groups.
5
Delete node
When you click this button, the node deletion dialogue will open. You will have to choose what to do with keywords located in the node – the system can either delete them or move to "Non-grouped".
6
Rename node
The function for changing node names.
7
Pin folder
Clicking this button marks the folder pin checkbox. This may be useful when searching by folders. For example, you can pin the "Recipes" node and specify "dough+dumplings" in the search, this will leave visible the Recipes node and all nodes with "dumplings" and "dough" text in them. This will allow you to transfer all articles about dough for dumplings into the Recipes folder.
8
General settings

This allows to manage display of folder parameters in the category tree. The following parameters are available for display:

  • Keyword
  • Total number of keywords
  • Search volume of the top sv keyword
  • Search volume of the top [!sv] keyword
  • Total search volume sv
  • Total search volume [!sv]
Most functions of the category tree are duplicated in the context menu. To call a folder's context menu, right-click its name.
Keywords area
The keywords area is located to the right. Into this area keywords, related to a node selected in the left part of the screen, are downloaded. I.e. if you go to the "Non-grouped", a list of keywords not associated with tree nodes will be displayed.
Main keyword of a folder is separately marked in the keywords table, you also can change main keywords. To do so, right-click a row you need in the table, and select "Make the main keyword" in the context menu that will open.

The keywords area has filtering, search and sorting functions, similar to functions in any other report on keys.so. You can get a list of keywords as a CSV file.

The following data is specified for each keywords: its text, broad search volume, and search volume "[!in !this !exact !format]", i.e. with this fixed set of words, their order and forms.

Above the list of keywords the report management button block is located.
Report management
1
Clusterer
Launches the function of keywords clustering, i.e. automatic sorting of keywords into groups;
2
Add keywords
The functional block for adding keywords to a project. If a keyword being added is already in the report, it will be ignored. This block has the following functions:
- From a list: adding keywords by specifying them in a text field;
- From a file: adding keywords from a text file (convenient with large numbers of keywords);
- From a group report: adding keywords from any of your group reports.
3
Delete keywords
Functions for deleting undesirable keywords from a report. The following options for this operation are provided here:
- According to a list: provide a list of keywords, and all of them will be removed from your project;
- According to a file: similarly, but with specifying keywords in a text file;
- According to a filter: You can specify which keywords are of no interest to you, for example keywords with low search volume;
- According to a group filter: You can delete groups if they fail to meet some conditions, for example those with too low or too high number of keywords, with search volume of their main keyword above or below a threshold, with total search volume of all keywords in a group above or below a required value. When deleting you can specify what to do with keywords – place them into the "Deleted" folder or the "Non-grouped" folder. You also can specify a method of filtering – whether all specified filters should be applied at once, or whether meeting any of the conditions is sufficient.
- Delete inexact duplicates. The system will analyze all keywords in a project and out of all inexact duplicates it will leave only the one that has the highest exact search volume. I.e. all the "house purchase" and "purchase house" permutations will be reduced to a single form.
- According to a blacklist. You can specify a list of undesirable matches, all keyword containing them will be deleted. For example such matches as "free download", "do it yourself, or "official website".

4
Export
Allows to turn a generated semantic kernel into a CSV file for further use.
Creating a group/Moving keywords according to a filter
Creating a new keyword folder with marked keywords or with a group of keywords moved according to the current filter in the keyword table.
Specifying settings for creating a new group according to a filter.
To move keywords into a new folder, you need to specify its name, parent folder, and filtering parameters: only selected keywords, or those which meet conditions of a filter in the keyword table.

To move keywords into an existing folder, you need to select a folder in the category tree and filtering parameters: only selected keywords, or those which meet conditions of a filter in the keyword table.
Clustering of keywords
After clicking the "Clusterer" button, the form for specifying parameters of algorithm operation opens.
Specifying clustering algorithm settings
By moving the slider you can control strength of keyword linking. The higher percentage you specify, the smaller groups will be, with more closely linked keywords. If you lower the threshold, groups formed will be more general, usually with more keywords in each.

You can specify the "Minimum [!search volume] of a group's main keyword" option. If this checkbox is checked, you will be able to specify the minimum acceptable search volume for the group's main keyword. I.e., the group will not form until it includes at least one keyword with the specified search volume.

You can specify the "Minimum total [!search volume] of a group" option. If this checkbox is checked, you will be able to specify the minimum acceptable total search volume for all keywords in the group. I.e., the group will not form until its total search volume reaches a specified value.

After clicking the "Start" button, the system will start the clustering process.

After clustering finishes, another service note will be created in the tree – Clustering result. I will include all nodes which your algorithm managed to collect automatically. Keywords not included into any of the group will remain in the "Non-grouped" node. In this case, the algorithm with default setting managed to sort into numerous groups 47,485 keywords out of 51,566.
How to work with groups?
Groups can be created, deleted, and moved between different nodes. Our example has a semantic kernel regarding citizenship and visas, so let's create the "Europe" node, and in it the "Greece" node. Now pin the "Greece" mode. Enter the "greec" filter and press Enter, all nodes which names include this text will display.

Highlight the first node, hold Shift and click the last node, this will select all of the nodes. You can also use the Ctrl key, in this case each click will select the node you clicked.

Now your can move all the nodes into your freshly created folder for Greece-related nodes. Enter the "athen" or "greec" filter and generally try to find more nodes which still may be related to the selected country, drag everything you've found to your "Greece" node.
How to work with keywords?
Select any node, the list of keywords associated with this node will be downloaded on the right. You can select or deselect any keyword in the table by clicking its row. To the left of the checkbox is the "hamburger" icon, by pulling it you can move the entire list of highlighted keywords. Drag and drop all keywords into any other node, and they will be associated with their new owner. Also when selecting keywords the function toolbar with three buttons will appear in the lower right part:

1. To Non-grouped – moves selected keywords to the "Non-grouped" keyword group;
2. To Deleted – deletes selected keywords;
3. Clear – clears selection.
Good luck!
We tried to make our tool maximally useful
After finishing operation you can download your semantic kernel as a CSV file. To do this click the "Export" button. You can specify keyword sorting options within nodes, and, optionally, remove from the file the "Non-grouped" service group of keywords.