Intersection

This module provide some method of PSI(Private Set Intersection)

RSA Intersection

This mode implements algorithm based on [RSA Intersection]. This work is built on FATE, eggroll and federation API that construct the secure, distributed and parallel infrastructure.

Our Intersection module tries to solve Privacy-Preserving Entity Match problem. This module helps two and more parties to find common entry ids without leaking non-overlapping ids. The process is illustrated below in figure 1.

../../../../../_images/rsa_intersection.png

Figure 1 (RSA Intersection between party A and party B)

In figure 1 ,Party A has user id u1,u2,u3,u4, while Party B has u1,u2,u3,u5. After Intersection, party A and party B both learn their common user ids, which are u1,u2,u3, while neither party A nor B could decrypt each other’s non-overlapping user ids. Transmission parties’ processed id to the other party, like \(Y-A\) and \(Z-B\), will not reveal raw ids. Processed \(Z-B\) is safe due to the privacy key of party B. Each \(Y-A\) includes different random value which binds to each value in \(X-A\) and will be safe as well.

Introduced in FATE version 1.6, split_calculation option is available for improved efficiency. Different from unified process described above, split_calculation process first splits hash-processed ids into even and odd groups; each group then runs through the RSA intersection process with either host or guest as the joining role. Note than with split_calculation, host(s) always know about their common even ids with guest since they are responsible for finding common even ids.

With RSA intersection, participants can get their intersection ids securely and efficiently.

RAW Intersection

This intersection mode implements the simple intersection method in which a participant sends all its ids to another participant, and the other participant finds their common ids. Finally, the joining role will send the intersection ids to the sender.

Multi-Host Intersection

Both RSA and RAW intersection support multi-host scenario. It means a guest can perform intersection with more than one host simultaneously and get the common ids among all participants.

../../../../../_images/multi_hosts.png

Figure 2 (multi-hosts Intersection)

Refer to figure 2 for a demonstration of one guest running intersection with two hosts; the same process applies to cases with more than two hosts. First, guest will run intersection with each host and get respective overlapping ids. Then, guest will find common IDs from all intersection results. Optionally, guest will send common IDs to every host.

Repeated ID intersection

We support repeated id intersection for some applications. In this case, one should provide the mask id to be mapped to repeated ids. For instances, in Guest, your data is

mask_id, id, value
alice_1, alice, 2
alice_2, alice, 3
bob, bob, 4

In Host, you data is

id, value
alice, 5
bob, 6

After intersection, guest will get the intersection results:

mask_id, value
alice_1, 2
alice_2, 3
bob, 4

And in host:

id, value
alice_1, 5
alice_2, 5
bob, 4

Set parameter repeated_id_process to true if you wish to use this mapping function for repeated ids.

Param

class EncodeParam(salt='', encode_method='none', base64=False)

Define the hash method for raw intersect method

Parameters

  • salt (the src data string will be str = str + salt, default by empty string) –

  • encode_method (str, the hash method of src data string, it support md5, sha1, sha224, sha256, sha384, sha512, sm3, default by None) –

  • base64 (bool, if True, the result of hash will be changed to base64, default by False) –

class IntersectParam(intersect_method: str = 'raw', random_bit=128, sync_intersect_ids=True, join_role='guest', with_encode=False, only_output_key=False, encode_params=<federatedml.param.intersect_param.EncodeParam object>, rsa_params=<federatedml.param.intersect_param.RSAParam object>, intersect_cache_param=<federatedml.param.intersect_param.IntersectCache object>, repeated_id_process=False, repeated_id_owner='guest', with_sample_id=False, allow_info_share: bool = False, info_owner='guest')

Define the intersect method

Parameters

  • intersect_method (str, it supports 'rsa' and 'raw', default by 'raw') –

  • random_bit (positive int, it will define the encrypt length of rsa algorithm. It effective only for intersect_method is rsa) –

  • sync_intersect_ids (bool. In rsa, 'synchronize_intersect_ids' is True means guest or host will send intersect results to the others, and False will not.) – while in raw, ‘synchronize_intersect_ids’ is True means the role of “join_role” will send intersect results and the others will get them. Default by True.

  • join_role (str, role who joins ids, supports "guest" and "host" only and effective only for raw. If it is "guest", the host will send its ids to guest and find the intersection of) – ids in guest; if it is “host”, the guest will send its ids to host. Default by “guest”.

  • with_encode (bool, if True, it will use hash method for intersect ids. Effective only for "raw".) –

  • encode_params (EncodeParam, it effective only for with_encode is True) –

  • rsa_params (RSAParam, effective for rsa method only) –

  • only_output_key (bool, if false, the results of intersection will include key and value which from input data; if true, it will just include key from input) – data and the value will be empty or some useless character like “intersect_id”

  • repeated_id_process (bool, if true, intersection will process the ids which can be repeatable) –

  • repeated_id_owner (str, which role has the repeated ids) –

  • with_sample_id (bool, data with sample id or not, default False; set this param to True may lead to unexpected behavior) –

class RSAParam(salt='', hash_method='sha256', final_hash_method='sha256', split_calculation=False, random_base_fraction=None, key_length=1024)

Define the hash method for RSA intersect method

Parameters

  • salt (the src data string will be str = str + salt, default '') –

  • hash_method (str, the hash method of src data string, it support sha256, sha384, sha512, sm3, default sha256) –

  • final_hash_method (str, the hash method of result data string, it support md5, sha1, sha224, sha256, sha384, sha512, sm3, default sha256) –

  • split_calculation (bool, if True, Host & Guest split operations for faster performance, recommended on large data set) –

  • random_base_fraction (positive float, if not None, generate (fraction * public key id count) of r for encryption and reuse generated r;) – note that value greater than 0.99 will be taken as 1, and value less than 0.01 will be rounded up to 0.01

  • key_length (positive int, bit count of rsa key, default 1024) –

Feature

Both RSA and RAW intersection support:

  1. Multi-host modeling task. The detailed configuration for multi-host modeling task is located here.

  2. Repeated ID intersection using ID expanding.

  3. Configurable hashing methods, including sha256, md5, and sm3; hash operators of RSA intersection can be configured separately, please refer here for more details.

RAW intersection supports the following extra feature:

  1. base64 encoding may be used for all hashing methods.