# US 2021/0241073 A1 · Ai-based keywork predictions for titles > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2021/0241073 A1 - Document ID: US20210241073A1 - Status: Published application - Kind: A1 - Application number: 16/782,945 - Filed: 2020-02-05 - Published: 2021-08-05 - Earliest claimed: 2020-02-05 - Assignees: eBay Inc. - Applicants: eBay Inc. - Inventors: Gilad Eliyahu Fuchs; Yoni Acriche - Web page: https://patentgrep.com/patent/US20210241073A1 ## CPC - G06F 16/345: {Summarisation for human users} - G06N 3/042: Knowledge-based neural networks; Logical representations of neural networks - G06F 40/166: Editing, e.g. inserting or deleting - G06N 3/08: Learning methods - G06N 3/09: Supervised learning - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Systems and methods for managing keyword predictions for proposed titles are provided. In example embodiments, a network system receives, from a user during a publication creation process, a proposed title for a publication associated with an item. The proposed title includes a plurality of tokens, whereby the plurality of tokens comprises at least all non-stock words in the proposed title. Based on the proposed title, the network system identifies an importance of each token of the plurality of tokens in the proposed title. The network system then causes presentation of a user interface that visually indicates the importance of each token of the plurality of tokens in the proposed title. ## Claims ### Claim 1 A method comprising: receiving, by a network system, a proposed title for a publication from a user during a publication creation process, the proposed title comprising a plurality of tokens and the publication being associated with an item, the plurality of tokens comprising at least all non-stock words in the proposed title; based on the proposed title, identifying, using a hardware processor of the network system, an importance of each token of the plurality of tokens in the proposed title; and causing presentation, by the network system, of a user interface that visually indicates the importance of each token of the plurality of tokens in the proposed title. ### Claim 2 (depends on 1) The method of claim 1, wherein the importance of a token is based on a probability that other users will include the token in their search query for similar items to the item. ### Claim 3 (depends on 1) The method of claim 1, wherein the causing presentation comprises presenting a higher importance token in a larger size than a lower importance token. ### Claim 4 (depends on 1) The method of claim 1, wherein the causing presentation comprises including a recommendation to remove a token with low importance from the proposed title. ### Claim 5 (depends on 1) The method of claim 1, wherein the causing presentation comprises including a recommendation to add a new token to the proposed title. ### Claim 6 (depends on 5) The method of claim 5, further comprising: receiving an indication to add the new token to the proposed tide; and updating the importance of each token in the proposed title based on the addition of the new token. ### Claim 7 (depends on 1) The method of claim 1, wherein the identifying the importance comprises using a prediction model to predict a probability for each token based on likelihood that other users will include each of the tokens in their search query for items similar to the item, the importance being based on the probability. ### Claim 8 (depends on 7) The method of claim 7, further comprising generating the prediction model, the generating comprising: collecting historical data of search queries and titles selected in response to the search queries, a search query and title selected in response to the search query forming a data pair; and training the prediction model using the data pair, the training including learning a structure of the tokens in the titles of the data pair using a neural network. ### Claim 9 A system comprising: one or more hardware processors; and a memory storing instructions that, when executed by the one or more hardware processors, causes the one or more hardware processors to perform operations comprising: receiving, by a network system, a proposed title for a publication from a user during a publication creation process, the proposed title comprising a plurality of tokens and the publication being associated with an item, the plurality of tokens comprising at least all non-stock words in the proposed title; based on the proposed title, identifying an importance of each token of the plurality of tokens in the proposed title; and causing presentation, by the network system, of a user interface that visually indicates the importance of each token of the plurality of tokens in the proposed title. ### Claim 10 (depends on 9) The system of claim 9, wherein the importance of a token is based on a probability that other users will include the token in their search query for similar items to the item. ### Claim 11 (depends on 9) The system of claim 9, wherein the causing presentation comprises presenting a higher importance token in a larger size than a lower importance token. ### Claim 12 (depends on 9) The system of claim 9, wherein the causing presentation comprises including a recommendation to remove a token with low importance from the proposed title. ### Claim 13 (depends on 9) The system of claim 9, wherein the causing presentation comprises including a recommendation to add a new token to the proposed title. ### Claim 14 (depends on 13) The system of claim 13, wherein the operations further comprise: receiving an indication to add the new token to the proposed title; and updating the importance of each token in the proposed title based on the addition of the new token. ### Claim 15 (depends on 9) The system of claim 9, wherein the identifying the importance comprises using a prediction model to predict a probability for each token based on likelihood that other users will include each of the tokens in their search query for items similar to the item, the importance being based on the probability. ### Claim 16 (depends on 15) The system of claim 15, wherein the operations further comprise generating the prediction model, the generating comprising: collecting historical data of search queries and titles selected in response to the search queries, a search query and title selected in response to the search query forming a data pair; and training the prediction model using the data pair, the training including learning a structure of the tokens in the titles of the data pair using a neural network. ### Claim 17 A machine-storage medium storing instructions that, when executed by one or more processors of a machine, cause the one or more processors to perform operations comprising: receiving, by a network system, a proposed title for a publication from a user during a publication creation process, the proposed tide comprising a plurality of tokens and the publication being associated with an item, the plurality of tokens comprising at least all non-stock words in the proposed title; based on the proposed title, identifying an importance of each token of the plurality of tokens in the proposed title; and causing presentation, by the network system, of a user interface that visually indicates the importance of each token of the plurality of tokens in the proposed title. ### Claim 18 (depends on 17) The machine-storage medium of claim 17, wherein the causing presentation comprises including a recommendation to remove a token with low importance from the proposed title. ### Claim 19 (depends on 17) The machine-storage medium of claim 17, wherein the causing presentation comprises including a recommendation to add a new token to the proposed title. ### Claim 20 (depends on 17) The machine-storage medium of claim 17, wherein the identifying the importance comprises using a prediction model to predict a probability for each token based on likelihood that other users will include each of the tokens in their search query for items similar to the item, the importance being based on the probability.