“Any sufficiently advanced technology is indistinguishable from magic”
Arthur C. Clarke
Phonetic Trademark search is very useful if one’s trademark doesn’t pull out any relevant results in the wordmark search. Although this search is highly advisable to run in addition to the wordmark search type. With its useful results, one can select a new trademark with the minimum possibility of facing S.11 objection or opposition from a third party on the basis of its similarity with any registered mark.
For instance, if an applicant comes up with ‘REALLEAF’ word to register for a specific class of goods but there’s an existing trademark ‘RELIEF’ registered in the same class it would not be safe from any legal trouble to go for it.
Technically, the databases give search results on the basis of Vector Space Model algorithms. They are trained to cluster similar trademark in spelling as well as in pronunciation by encoding words & sentences as numeric vectors. This vector aims to capture the semantic properties of the word – words whose vectors are close together should be similar in terms of semantic meaning. In this model, a user enters a short free-text query, and documents are ranked based on their similarity to the query. The model assumes that the relevance of a document to a query is roughly equal to the document-query similarity. Both the documents and queries are represented using the bag-of-words model.
For example, consider a query “new new times”. A very small collection ‘C’ consists in the following documents. According to calculation of its vector, scores & the similarity values, the order in which the documents will be presented as result to the query will be
d1: “new york times”
d2: “new york post”
d3: “los angeles times”
Search incapability
Despite the Trademark public search database being efficient there continue to be flaws in this facility. Under the Phonetic type of search, the database is unable to actually compare the trademarks matching phonetically. The results are a mixed bag with a lot of useless results.
Following are some hypothetical examples:
- When trademark ‘THE HERBIS’ is searched in class 5 as a fresh trademark you get 812 search results. TRIVAX, TRIOPAQUE, TROPHYSAN, TRIVISOL, TRAVASE are the first five and the list goes the same way. These names are undoubtedly and undeniably irrelevant to the searched name. They appear due to the sound of the letter ‘R’ but that doesn’t make it sound similar to ‘THE HERBIS’ in any way. There might be a few similar names in this unending list but to figure that out one has to go through ALL the names.
- When ‘LaCarl’ is searched in class 25 the search result shows LUXER, LIGER, LEEZAR, LAUGH & WEAR, LEISURE in the list which does not match with LaCarl at all. The trademarks that can match are LA-CUER, LE CARLO, LEUCARLO. But to find out the matching ones, one has to check all the marks that appear in its search result.
Further, this kind of enormously unnecessary data becomes a challenge for a fresh applicant to pinpoint the trademarks that are similar to his trademark. As the search results are absolutely dissimilar to the searched name there arises no question of their conflict with it. The challenge of going through so much of data cause wastage of time and brings the applicant’s trademark registration at the risk of getting alleged as an infringement or passing off from business competitors.
Conclusion
The Trademark Amendment Rules, 2017 brought advancements in the e-filing system yet this problem continues to exist. Thus, it is highly recommended that the system should be updated to facilitate applicants finding out conflicting trademarks to ensure their trademark safely proceeds toward registration. This would help in reducing the number of objections by the trademark Registry as well as reducing the excess amount of show cause and opposition hearings which has become a burden on the Registrar of trademarks.