Normalizing non-numeric features of files

• US 10,078,666 B2
• Filed: 11/05/2015
• Issued: 09/18/2018
• Est. Priority Date: 11/28/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Li, Chang Sheng, Meng, Fan Jing, Stern, Edith Helen, Wang, Han, Xu, Jing Min, Yang, Lin, Zhuo, Xuejun
• Primary Examiner(s): Kim, Taelor
• Assistant Examiner(s): Frumkin, Jesse P
• Application Number:
  US14/933,382
• Publication Number:
  US 20160154801A1
• Time in Patent Office: 1,048 Days
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: H04L 41/0806 for initial configuration or provisioning, e.g. plug-and-play

Abstract

Embodiments include method, computer program products and apparatuses for normalizing non-numeric features of files and corresponding apparatus. Aspects include segmenting at least one pair of positive instances of a non-numeric feature of a file into a number of tokens and comparing the tokens in the at least one pair of positive instances to obtain matching tokens. Aspects also include calculating weights of their matching the file, for the matching tokens, and storing the tokens and their weights in a token base.

Claims

1. An apparatus for normalizing non-numeric features of files, comprising:

a token segmenting module configured to segment at least one pair of positive instances of a non-numeric feature of a file into a number of tokens, wherein the non-numeric feature of the file comprises a file storage path of a configuration file stored in a networked computer environment;

a token matching module configured to compare the tokens in the at least one pair of positive instances to obtain matching tokens, wherein the token matching module comprises;

a token matching score calculating sub-module configured to calculate a maximum matching score between each token in a positive instance with the tokens in another positive instance;

a token selecting sub-module configured to select the tokens of which the maximum matching scores are greater than a given threshold, to get the matching tokens; and

a token base constructing module configured to, for the matching tokens, calculate weights of their matching the file, and store the tokens and their weights in a token base, wherein the matching tokens identify similar configuration files stored in different locations in the networked computer environment.

2. The apparatus of claim 1, further comprising:

a token sequence forming module configured to form token sequences from the matching tokens as segments of the positive instances; and

the token base constructing module further configured to, for the token sequences, calculate weights of their matching the file, and store the token sequences and their weights in the token base.

3. The apparatus of claim 2, further comprising:

the token segmenting module further configured to, in response to receiving a test instance of the non-numeric feature of a test file, segment the test instance into a number of tokens and token sequences;

a normalizing module configured to calculate a score that the test instance matches the file by matching the tokens and the token sequences in the test instance with the tokens and token sequences in the token base, as a normalized numeric feature of the non-numeric feature of the test file.

4. The apparatus of claim 3, wherein the normalizing module comprises:

a token and sequence combining sub-module configured to obtain different combinations formed by tokens and token sequences of the test instance;

a token base comparing sub-module configured to compare each token and token sequence in each combination with the tokens and token sequences in the token base, to get the most similar token or token sequence in the token base, and a similarity and the weight thereof;

a weight calculating sub-module configured to, for each combination, according to the similarities and weights of the most similar tokens or token sequences of its tokens and token sequences, calculate a weight of the combination; and

a maximum weight selecting sub-module configured to select the maximum weight in the different combinations, as the score that the test instance matches the file.

5. The apparatus of any one of claim 3, further comprising:

a determining module configured to determine whether the test file is a given file according to the normalized numeric feature of the test file.

6. The apparatus of claim 2, wherein the segmenting the positive instance into a number of tokens is performed by using structural characteristics of the non-numeric feature.

7. The apparatus of claim 1, wherein the token base constructing module comprises:

a weight calculating sub-module configured to, according to at least one of the number of appearances of the matching token or token sequence in the group of positive instance, the token length of the token or the token sequence, and the token position of the token or the token sequence in the positive instance, calculate a weight of its matching the file, wherein the greater the number of appearances is, the greater the token length is or the more backward the token position is, the greater the weight is.

8. A computer program product for normalizing non-numeric features of files, the computer program product comprising:

a non-transitory storage medium readable by a processing circuit and storing instructions for execution by the processing circuit for performing a method comprising;

segmenting at least one pair of positive instances of a non-numeric feature of a file into a number of tokens, wherein the non-numeric feature of the file comprises a file storage path of a configuration file stored in a networked computer environment;

comparing the tokens in the at least one pair of positive instances to obtain matching tokens, wherein the comparing comprises;

calculating a maximum matching score between each token in a positive instance with the tokens in another positive instance;

selecting the tokens of which the maximum matching scores are greater than a given threshold, to get the matching tokens; and

for each of the matching tokens, calculating weights of their matching the file, and storing the tokens and their weights in a token base, wherein the matching tokens identify similar configuration files stored in different locations in the networked computer environment.

9. The computer program product method of claim 8, further comprising:

forming token sequences from the matching tokens as segments of the positive instances; and

for each of the token sequences, calculating weights of their matching the file, and storing the token sequences and their weights in the token base.

10. The computer program product of claim 9, further comprising:

in response to receiving a test instance of the non-numeric feature of a test file, segmenting the test instance into a number of tokens and token sequences;

calculating a score that the test instance matches the file by matching the tokens and the token sequences in the test instance with the tokens and token sequences in the token base, as a normalized numeric feature of the non-numeric feature of the test file.