Determining collaboration recommendations from file path information

Patent metadata

Field	Value
Publication number	US-11928083-B2
Application number	US-201916264357-A
Country	US
Kind code	B2
Filing date	Jan 31, 2019
Priority date	Oct 9, 2017
Publication date	Mar 12, 2024
Grant date	Mar 12, 2024

Abstract

Methods, systems and computer program products for recommendation systems. Embodiments commence by gathering a set of pathnames that refer to content objects of a collaboration system. A tokenizer converts at least some of the pathnames into vectors. The vectors comprise hierarchical path components such as folder names or file names, which vectors are labeled with an indication as to whether or not the folder or file referred to in a particular vector had been clicked on by one or more users. Some portion of the labeled vectors are used to train a predictive model. Another portion of the vectors are used to validate the predictive model. When the model exhibits sufficient precision and recall, the predictive model is then used to predict the probability that a particular user would have an interest in a particular folder or file. The folder name or file name is presented as a collaboration recommendation.

First claim

What is claimed is:

1. A method for determining collaboration recommendations from file path information, the method comprising:
gathering a set of pathnames of content objects in a collaboration platform;
converting at least some of the set of pathnames of the content objects into vectors, wherein
a vector includes both a directory hierarchy representation and an interaction attribute,
the vectors are stored in a feature vector data structure,
the feature vector data structure comprises a data record describing an object identifier of a content object of the content objects, a list of one or more users interacting with the content object, information of an interaction type for a pair of the content object and a user of the list of one or more users, and one or more attributes pertaining to the directory hierarchy representation,
the directory hierarchy representation stored in the vector comprises a first file path token and a second file path token,
the first file path token comprises at least a first name string representing a first hierarchy of a pathname associated with a directory hierarchy having multiple levels,
the second file path token comprises at least a second name string representing a second hierarchy of the pathname associated with the directory hierarchy,
the first hierarchy is located at a different level of the multiple levels from the second hierarchy in the directory hierarchy representation, and
the interaction attribute represents an interaction event that corresponds to an operation that is observed to have been performed by a specific user or by a system for the specific user over a specific content object to indicate presence or absence of the interaction event between the specific user and the specific content object;

generating a predictive model from at least some of the vectors and respective interaction attributes; and
providing a recommended content object at least by providing an input pertaining to a plurality of content objects to the predictive model and at least by applying the predictive model to the input to determine one or more respective statistic measures pertaining to the plurality of content objects, wherein a collaboration recommendation pertaining to the recommended content object comprises at least one of a folder or a file.

2. The method of claim 1 , further comprising:
recording one or more interaction attributes that correspond to interaction events between users or operations performed by at least one of the users or by the system over one or more content objects of the plurality of content objects.

3. The method of claim 2 , wherein the predictive model is generated based at least in part on the one or more interaction attributes.

4. The method of claim 1 , wherein
the directory hierarchy representation is codified in the vector,
the pathname having a form “/A/B/” for the specific content object is tokenized such that a first sub-directory of the directory hierarchy “A” forms the first file path token, and
a second sub-directory of the directory hierarchy “B” forms the second file path token of the vector for the specific content object.

5. The method of claim 4 , wherein
the vector is accessed to determine one or more recommended content objects comprising the recommended content object at least by determining, at the predicted model, one or more respective click probabilities of the one or more recommended content objects based at least in part upon the one or more attributes of respective pathnames of the one or more recommended content objects, and
information pertaining to the one or more recommended content objects is ordered based at least in part upon the one or more respective click probabilities.

6. The method of claim 5 , further comprising:
processing the one or more recommended content objects, wherein processing the one or more recommended content objects comprises:
determining a display area in a user interface of a user device for displaying at least the recommended content object; and
filtering the one or more recommended content objects based at least in part upon the display area.

7. The method of claim 6 , wherein processing the one or more recommended content objects includes at least one of scoring, sorting, ranking, filtering, decorating, or presenting the one or more recommended content objects.

8. The method of claim 7 , further comprising:
constructing one or more recommendation messages based at least in part on the one or more recommended content objects; and
presenting the one or more recommendation messages at the user interface based at least in part upon the one or more respective click probabilities, wherein a same file in multiple different folders having multiple different pathnames have different click probabilities.

9. A non-transitory computer readable medium having stored thereon a sequence of instructions which, when stored in memory and executed by one or more processors causes the one or more processors to perform a set of acts for determining collaboration recommendations from file path information, the set of acts comprising:
gathering a set of pathnames of content objects in a collaboration platform;
converting at least some of the set of pathnames of the content objects into vectors, wherein
a vector includes both a directory hierarchy representation and an interaction attribute,
the vectors are stored in a feature vector data structure,
the feature vector data structure comprises a data record describing an object identifier of a content object of the content objects, a list of one or more users interacting with the content object, information of an interaction type for a pair of the content object and a user of the list of one or more users, and one or more attributes pertaining to the directory hierarchy representation,
the directory hierarchy representation stored in the vector comprises a first file path token and a second file path token,
the first file path token comprises at least a first name string representing a first hierarchy of a pathname associated with a directory hierarchy having multiple levels, the second file path token comprises at least a second name string representing a second hierarchy of the pathname associated with the directory hierarchy,
the first hierarchy is located at a different level of the multiple levels from the second hierarchy in the directory hierarchy representation, and
the interaction attribute represents an interaction event that corresponds to an operation that is observed to have been performed by a specific user or by a system for the specific user over a specific content object to indicate presence or absence of the interaction event between the specific user and the specific content object;

generating a predictive model from at least some of the vectors and respective interaction attributes; and
providing a recommended content object at least by providing an input pertaining to a plurality of content objects to the predictive model and at least by applying the predictive model to the input to determine one or more respective statistic measures pertaining to the plurality of content objects, wherein a collaboration recommendation pertaining to the recommended content object comprises at least one of a folder or a file.

10. The non-transitory computer readable medium of claim 9 , further comprising instructions which, when stored in memory and executed by the one or more processors causes the one or more processors to perform acts of:
recording one or more interaction attributes that correspond to interaction events between users or operations performed by the specific user or by the system over one or more content objects.

Assignees

• Box Inc

Inventors

• Vengerov David
• Jalagam Sesh
• De Vansa Vikramaratne Victor

Classifications

• G06F16/176 Primary

  Support for shared access to files; File sharing support · CPC title

• G06F16/14 Primary

  Details of searching files based on file metadata · CPC title

• G06F16/156

  Query results presentation · CPC title

• G06F16/166

  File name conversion · CPC title

• G06F21/6209

  to a single file or object, e.g. in a secure envelope, encrypted and accessed using a key, or with access control rules appended to the object itself · CPC title